| jroll | yes, just distracted | 00:00 |
|---|---|---|
| NobodyCam | :) | 00:00 |
| jroll | devananda: also, what about dmitry's concerns? | 00:00 |
| *** naohirot has joined #openstack-ironic | 00:01 | |
| openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add fallback secure erase using shred https://review.openstack.org/165247 | 00:04 |
| * jroll bbiab | 00:05 | |
| openstackgerrit | Josh Gachnang proposed openstack/ironic-python-agent: Add cleaning/zapping support to IPA https://review.openstack.org/161066 | 00:07 |
| BadCub | I added 165247 & 161066 to the EtherPad under Cleaning | 00:08 |
| devananda | JoshNang: is https://bugs.launchpad.net/ironic/+bug/1409169 really a High Pri bug? | 00:08 |
| openstack | Launchpad bug 1409169 in Ironic "IPA LLDP lookups loop forever" [High,Triaged] - Assigned to Ricardo Ferreira (rsff) | 00:08 |
| devananda | jroll: oh. i'm lame and mixing up patches. clearly time to take a break .... | 00:09 |
| JoshNang | devananda: definitely not. low priority for sure (it's been running successfully every decom for months, only had one instance of it freaking out) | 00:09 |
| devananda | NobodyCam: didn't you work on something similar to https://bugs.launchpad.net/ironic/+bug/1403887 ? | 00:11 |
| openstack | Launchpad bug 1403887 in Ironic "No such file or directory error during deployment" [High,Triaged] | 00:11 |
| devananda | if that's really a High bug why aren't we hitting it all the time? | 00:11 |
| JayF | devananda: the specific result of #1409169: If you have some problem with network (in our case; it was a bad NIC cable), node is stuck in CLEANING ... but there are no upstream managers that use those methods, so it's probably fine | 00:12 |
| JayF | one of those things where priority vs severity are radically different | 00:12 |
| devananda | yea | 00:12 |
| JayF | it's really severe but low priority | 00:12 |
| * devananda hates LP | 00:13 | |
| * BadCub hate LP more LOL | 00:13 | |
| JoshNang | JayF: even then, we only hit it once. the other nodes with bad cables/ports/etc threw errors appropriately | 00:13 |
| JayF | Really? Man that server really rolled a 1 then, eh? | 00:13 |
| devananda | JayF: and even then, can't you work around that by power cycling the node or something? | 00:13 |
| JoshNang | JayF: hehe | 00:13 |
| JayF | devananda: workaround is nbd, discovery can be difficult if you've got a big cluster and things constantly cycling | 00:14 |
| devananda | high == serious impact in functionality | 00:14 |
| devananda | see https://wiki.openstack.org/wiki/BugTriage#Task_2:_Prioritize_confirmed_bugs_.28bug_supervisors.29 | 00:14 |
| JayF | So that one should be Medium? | 00:15 |
| devananda | yah | 00:15 |
| JayF | and in context, this would be in Ironic, not IPA context, right? | 00:15 |
| JayF | Like something could break IPA badly, but if it doesn't break Ironic badly it'd be lower rated | 00:15 |
| devananda | right | 00:15 |
| devananda | though ... | 00:15 |
| *** achanda has quit IRC | 00:15 | |
| devananda | once IPA is the default ramdisk that we gate with, that changes | 00:16 |
| JayF | I'll keep that somewhat in mind when filing bugs then, since we tend to use things downstream before upstream does :) | 00:16 |
| JayF | well, Ironic doesn't gate on LLDP helper functions in IPA | 00:16 |
| JayF | lol | 00:16 |
| devananda | exactly | 00:16 |
| devananda | i'd really say that's wishlist | 00:16 |
| devananda | because afaik there's no LLDP support upstream at all yet | 00:16 |
| *** achanda has joined #openstack-ironic | 00:16 | |
| JayF | Nothing upstream uses that code today | 00:16 |
| devananda | (but there are neat things in IPA that I hvaen't seen, i'm sure) | 00:16 |
| JayF | but downstream hardware managers can and do use it | 00:17 |
| JayF | I actually think that's an interesting thing to think about | 00:17 |
| devananda | JayF: so it's not a bug in Ironic, and it may be a medium bug in IPA | 00:17 |
| JayF | how much helpful hardware utility type functions should IPA have that aren't used directly in upstream hardware managers | 00:17 |
| JayF | devananda: well, IPA bugs are filed in Ironic, so tag:agent + Medium is really the only way to communicate that | 00:17 |
| *** oomichi has quit IRC | 00:17 | |
| JayF | maybe we need something like a tag:agent-only | 00:17 |
| JayF | to indicate it doesn't impact upstream default functionality | 00:18 |
| JayF | like if standalone mode broke, it'd be a bug in IPA we should fix but not something Ironic cares about | 00:18 |
| devananda | JayF: or a separate IPA project in LP .... as much as I dont want to separate the teams like that | 00:18 |
| *** achanda has quit IRC | 00:18 | |
| *** achanda has joined #openstack-ironic | 00:18 | |
| JayF | devananda: btw I need your +1s back on that governnace/project-config change | 00:18 |
| devananda | JayF: link? | 00:18 |
| JayF | devananda: I am ... surprised the process apparently requires governance change before p-c change | 00:18 |
| JayF | I actually think that changed while that patch was up | 00:19 |
| JayF | 3 useful and helpful +1s: https://review.openstack.org/#/c/155117/ https://review.openstack.org/#/c/155868/ https://review.openstack.org/#/c/164370/ | 00:19 |
| JayF | the first is making agent_ssh vote on devstack, which is unrelated but an easy review you should have input on anyway | 00:19 |
| *** ramineni has joined #openstack-ironic | 00:20 | |
| ramineni | NobodyCam: hi | 00:21 |
| BadCub | BRB | 00:22 |
| JoshNang | is there anyone with more experience in devstack that can help me find a solution here: https://review.openstack.org/#/c/164954/ | 00:22 |
| JayF | I'll take a look but not sure it'll be helpful | 00:23 |
| JoshNang | basically, i need a spot where neutron is started (so i can get the private network uuid) but before ironic is started (so it picks up the config change) | 00:23 |
| JayF | have you asked in #openstack-qa? | 00:23 |
| JoshNang | headed there now | 00:24 |
| *** mtanino has quit IRC | 00:30 | |
| *** achanda has quit IRC | 00:30 | |
| devananda | JoshNang: yea, -qa is the right place. jogo or adam_g or Shrews would be the first people I'd go to | 00:31 |
| JoshNang | devananda: thanks! they already gave me the pointer i needed | 00:31 |
| devananda | great | 00:31 |
| devananda | and with that - I am going to catch a bus to dinner | 00:31 |
| JoshNang | enjoy! | 00:31 |
| devananda | be back later, folks o/ | 00:32 |
| NobodyCam | night devananda | 00:34 |
| * BadCub needs to start dinner-y things. | 00:34 | |
| NobodyCam | hi ramineni :) | 00:35 |
| NobodyCam | I can not what my question was.. I hope I left a comment on the review :/ | 00:36 |
| NobodyCam | i can not THINK what... | 00:37 |
| NobodyCam | :-p | 00:37 |
| *** ChuckC has quit IRC | 00:53 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Common changes for secure boot support https://review.openstack.org/153974 | 00:54 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: ilo_iscsi driver do not validate boot_option https://review.openstack.org/164414 | 01:02 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Ilo drivers sets capabilities:boot_mode in node https://review.openstack.org/155731 | 01:03 |
| *** ramineni has quit IRC | 01:04 | |
| *** ramineni has joined #openstack-ironic | 01:13 | |
| ramineni | NobodyCam: :) | 01:13 |
| ramineni | NobodyCam: responded to your comments, please have a look | 01:14 |
| ramineni | NobodyCam: and ya, you have posted the comments :) https://review.openstack.org/#/c/157715/17 | 01:17 |
| *** saripurigopi has joined #openstack-ironic | 01:29 | |
| *** chenglch has joined #openstack-ironic | 01:37 | |
| *** saripurigopi has quit IRC | 01:46 | |
| *** zhenguo has quit IRC | 01:47 | |
| *** ramineni has quit IRC | 01:52 | |
| *** igordcard_ has quit IRC | 01:59 | |
| *** harlowja_ is now known as harlowja_away | 02:06 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 02:13 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 02:14 |
| *** coolsvap|afk is now known as coolsvap | 02:19 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for iscsi_ilo driver https://review.openstack.org/154814 | 02:26 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for agent_ilo driver https://review.openstack.org/154816 | 02:27 |
| jlvillal | rloo: I think Zuul is being obstinate: https://review.openstack.org/#/c/163572/ | 02:32 |
| jlvillal | rloo: It doesn't seem to have paid attention to your +A | 02:33 |
| rloo | jlvillal: that can't be merged until the dependent patch is merged. | 02:33 |
| jlvillal | rloo: Oh! :D | 02:33 |
| *** coolsvap is now known as coolsvap|afk | 02:33 | |
| rloo | jlvillal: :) | 02:34 |
| jlvillal | rloo: Thanks for explaining. | 02:34 |
| rloo | jlvillal: thx for noticing! | 02:34 |
| jlvillal | rloo: They did have one earlier today where they seemed to re-do the +A to get it to go. I thought the same thing had happened here. My mistake. | 02:35 |
| rloo | jlvillal: no worries. it could have been a problem, so worth calling out if you aren't sure. | 02:37 |
| jlvillal | rloo: Next time I'll look at the dependency :) | 02:37 |
| *** rloo has quit IRC | 02:44 | |
| *** coolsvap|afk is now known as coolsvap | 02:45 | |
| *** yuanying has quit IRC | 02:46 | |
| *** ramineni has joined #openstack-ironic | 02:53 | |
| *** lazy_prince has quit IRC | 02:57 | |
| *** killer_prince has joined #openstack-ironic | 03:04 | |
| *** killer_prince is now known as lazy_prince | 03:04 | |
| *** coolsvap is now known as coolsvap|afk | 03:09 | |
| *** Marga_ has quit IRC | 03:09 | |
| *** coolsvap|afk is now known as coolsvap | 03:16 | |
| BadCub | rloo thanks for the +A on 163572 :-) | 03:20 |
| *** yuanying has joined #openstack-ironic | 03:37 | |
| *** yuanying has quit IRC | 03:38 | |
| *** yuanying has joined #openstack-ironic | 03:38 | |
| *** saripurigopi has joined #openstack-ironic | 03:49 | |
| *** stendulker has joined #openstack-ironic | 04:00 | |
| *** ijw has quit IRC | 04:02 | |
| ramineni | JoshNang: hi, there? | 04:02 |
| *** kbs has quit IRC | 04:06 | |
| *** kbs has joined #openstack-ironic | 04:08 | |
| *** Marga_ has joined #openstack-ironic | 04:11 | |
| *** Marga__ has joined #openstack-ironic | 04:12 | |
| *** Marga_ has quit IRC | 04:16 | |
| openstackgerrit | Sirushti Murugesan proposed openstack/ironic: Raise exception for Agent Deploy driver when using partition images https://review.openstack.org/164440 | 04:18 |
| *** BadCub is now known as BadCub_Sleeping | 04:22 | |
| *** Haomeng has quit IRC | 04:23 | |
| *** rameshg87 has joined #openstack-ironic | 04:28 | |
| JoshNang | ramineni: hi! | 04:37 |
| ramineni | JoshNang:hi | 04:44 |
| rameshg87 | good morning ironic | 04:44 |
| JoshNang | good morning! | 04:45 |
| ramineni | JoshNang: morning :) have a question on cleaning for agent_driver patch | 04:45 |
| ramineni | JoshNang: sorry..almost night for you :P | 04:46 |
| JoshNang | heh oh definitely, almost 10 here | 04:46 |
| ramineni | JoshNang: The patch is not wrkng for me :( | 04:46 |
| JoshNang | ask away! i'm just about to push up the (hopefully final) patch | 04:46 |
| ramineni | JoshNang : throwing FailedToUpdateMacOnPort: Update MAC address on port: 65ed5871-7f74-4f7c-8d03-1f5d1612167a failed. | 04:46 |
| JoshNang | so i've run into that when i hit a cleanfail and tried to reclean without cleaning out ports | 04:47 |
| JoshNang | it creates ports with the node's macs, and doesn't properly clean them up. if you do a port-list and clean up new ports, you should be able to retry | 04:48 |
| ramineni | JoshNang: I have deleted corresponding neutron port for the mac . | 04:48 |
| JoshNang | (and i'm adding the tear down bits for when it fails) | 04:48 |
| ramineni | **rechecking | 04:48 |
| ramineni | JoshNang: but adding 'mac_address' in the body is not helping ? for creating neutron port with mac? | 04:49 |
| JoshNang | oh! | 04:50 |
| ramineni | JoshNang: but im getting , BadRequest: Cannot update read-only attribute mac_address | 04:52 |
| JoshNang | yeah, i couldn't get it to work with mac_address in the body | 04:52 |
| JoshNang | the last couple patch sets have that broken out into a separate call after creating the port. | 04:53 |
| JoshNang | (which is...frustrating, but i didn't have time to dig into why it's happening) | 04:53 |
| ramineni | JoshNang: ok .. i have the latest patch set downloaded , ill try freshly again ..removing ports | 04:54 |
| JoshNang | i've been testing this in devstack all day. that part has definitely been working | 04:54 |
| JoshNang | just about to test with again with JayF's 'shred' failsafe for erase_devices. vm's don't support secure erase and it was causing clean fails :/ | 04:55 |
| ramineni | JoshNang: let me check again , may be im missing smthng :( | 04:56 |
| JoshNang | :( it could be a problem with my code too | 04:56 |
| JoshNang | how are you testing? devstack? real hardware? | 04:56 |
| ramineni | devstack installed , but added the config paratmeter networkuuid | 04:58 |
| ramineni | Its definetly creating neutron port | bd48149e-e04c-4387-9bb8-5673ef5c5b62 | | fa:16:3e:ea:10:41 | {"subnet_id": "7ab91742-f052-4ab4-9edc-da126fa0e418", "ip_address": "10.10.1.210"} | | 04:58 |
| JoshNang | ahh yeah...we were fighting all day with that...should be fixed now | 04:58 |
| JoshNang | (the cleaning network uuid) | 04:58 |
| ramineni | but not able to update the mac with the one ironic port has | 04:59 |
| ramineni | stilll im getting on BadRequest: Cannot update read-only attribute mac_address :( | 04:59 |
| JoshNang | that's really weird. that was working fine went i went to sleep last night | 04:59 |
| JoshNang | to be clear, you code should not have body = {...'mac_address': foo}. | 05:00 |
| ramineni | it didnt have .. i removed | 05:00 |
| JoshNang | it should be something like self.update_port_address(port['port']['id'], ironic_port.address) | 05:00 |
| JoshNang | (in neutron.py) | 05:00 |
| ramineni | yes , | 05:01 |
| * ramineni debugging, why its failing for me | 05:02 | |
| *** penick has joined #openstack-ironic | 05:05 | |
| *** achanda has joined #openstack-ironic | 05:06 | |
| *** Haomeng has joined #openstack-ironic | 05:08 | |
| *** kbs1 has joined #openstack-ironic | 05:09 | |
| *** kbs1 has quit IRC | 05:09 | |
| *** kbs has quit IRC | 05:10 | |
| *** kbs has joined #openstack-ironic | 05:10 | |
| *** penick has quit IRC | 05:11 | |
| *** sdake_ has quit IRC | 05:13 | |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: iLO implementation for hardware inspection https://review.openstack.org/151596 | 05:19 |
| *** Nisha has joined #openstack-ironic | 05:20 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic: Address nits in uefi agent iscsi deploy commit https://review.openstack.org/165297 | 05:26 |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic: Address nits in uefi agent iscsi deploy commit https://review.openstack.org/165297 | 05:26 |
| ramineni | JoshNang: stangely, I could make it work with 'mac_address' in body rather than seperate function later | 05:27 |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: iLO implementation for hardware inspection https://review.openstack.org/151596 | 05:27 |
| JoshNang | ramineni: hmm. | 05:27 |
| JoshNang | if it works, i'd much rather do it in one call! | 05:27 |
| openstackgerrit | PSPradhan proposed stackforge/proliantutils: Update RIS library https://review.openstack.org/163290 | 05:31 |
| ramineni | JoshNang: ya | 05:31 |
| ramineni | JoshNang: not sure, why its not wrkng the one coded now , i updated the neutronclient and neutron also | 05:33 |
| JoshNang | so it is working with the mac address in body? | 05:34 |
| *** penick has joined #openstack-ironic | 05:35 | |
| ramineni | JoshNang: ya, it worked that way for me .. may be you could check again , if it wrks for you | 05:41 |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: iLO driver updates node capabilities during inspection https://review.openstack.org/163572 | 05:43 |
| *** subscope_ has joined #openstack-ironic | 05:44 | |
| *** saripurigopi has quit IRC | 05:59 | |
| jroll | JoshNang: holy cow you're up late | 06:10 |
| JoshNang | jroll: yuuup | 06:10 |
| JoshNang | almost done though :D | 06:11 |
| jroll | nice, well good evening | 06:11 |
| * jroll just got back from taking kelly to the airport | 06:11 | |
| JoshNang | ahh | 06:11 |
| jroll | forgot is was st paddy's, so many cops out | 06:11 |
| JoshNang | that makes sense. i'm usually up until midnight or so, but i don't wake up at 5 :P | 06:12 |
| jroll | :P | 06:12 |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic: Add pxe_ucs driver to manage Cisco UCS servers https://review.openstack.org/159734 | 06:18 |
| JoshNang | INFO ironic.conductor.manager [-] Node 0095c1b0-5dc8-48d5-9c1b-c56be951c43a cleaning complete | 06:19 |
| JoshNang | \o/ | 06:19 |
| mrda | \o/ | 06:19 |
| JoshNang | (with JayF's shred fix) | 06:19 |
| *** kalpase has joined #openstack-ironic | 06:20 | |
| *** kalpase has left #openstack-ironic | 06:20 | |
| JoshNang | ramineni: it didn't throw an error, but it the agent never finishes booting | 06:21 |
| JoshNang | oh, nvm. just got impatient. good call! updating the code. thanks! | 06:21 |
| *** saripurigopi has joined #openstack-ironic | 06:23 | |
| *** yog__ has joined #openstack-ironic | 06:25 | |
| *** achanda has quit IRC | 06:25 | |
| *** achanda has joined #openstack-ironic | 06:27 | |
| openstackgerrit | Josh Gachnang proposed openstack/ironic: Implement cleaning/zapping for the agent driver https://review.openstack.org/161453 | 06:28 |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: follow up patch for ilo capabilities https://review.openstack.org/165310 | 06:29 |
| JoshNang | ^ bam. works in devstack. almost certainly is going to fail tempest because a) the nova patch isn't in and deletes will time out b) the shred patch should land. cleaning is taking like 10 minutes to shred the whole disk :/ maybe we should shrink the disk or disable secure erase it for testing. | 06:29 |
| * JoshNang is going to sleep now | 06:30 | |
| rameshg87 | good night JoshNang | 06:31 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic: Implement cleaning/zapping for the agent driver https://review.openstack.org/161453 | 06:35 |
| openstackgerrit | Nisha Agarwal proposed stackforge/proliantutils: Adding functionality required by ilo-inspection https://review.openstack.org/164142 | 06:39 |
| *** ukalifon1 has joined #openstack-ironic | 06:44 | |
| openstackgerrit | Naohiro Tamura proposed openstack/ironic: Add pxe_irmc to the sending IPMI sensor data driver list https://review.openstack.org/165313 | 06:47 |
| *** subscope_ has quit IRC | 06:56 | |
| *** aswadr has joined #openstack-ironic | 07:14 | |
| *** Nisha has quit IRC | 07:17 | |
| *** pradipta has joined #openstack-ironic | 07:28 | |
| *** stendulker has quit IRC | 07:34 | |
| *** achanda has quit IRC | 07:35 | |
| *** achanda has joined #openstack-ironic | 07:39 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Common changes for secure boot support https://review.openstack.org/153974 | 07:42 |
| *** jcoufal has joined #openstack-ironic | 07:55 | |
| *** dlpartain has joined #openstack-ironic | 08:02 | |
| *** dlpartain has left #openstack-ironic | 08:02 | |
| *** pas-ha has joined #openstack-ironic | 08:08 | |
| *** achanda has quit IRC | 08:10 | |
| *** Marga__ has quit IRC | 08:13 | |
| *** pradipta has quit IRC | 08:14 | |
| *** hyakuhei has joined #openstack-ironic | 08:19 | |
| *** romcheg has joined #openstack-ironic | 08:20 | |
| *** jistr has joined #openstack-ironic | 08:22 | |
| *** hyakuhei has quit IRC | 08:25 | |
| *** lsmola has joined #openstack-ironic | 08:25 | |
| *** hyakuhei has joined #openstack-ironic | 08:25 | |
| *** yuanying has quit IRC | 08:27 | |
| *** dtantsur|afk is now known as dtantsur | 08:28 | |
| dtantsur | Morning | 08:28 |
| *** pradipta has joined #openstack-ironic | 08:30 | |
| *** ifarkas has joined #openstack-ironic | 08:30 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 08:32 |
| *** ndipanov has joined #openstack-ironic | 08:33 | |
| *** uschreiber_ has joined #openstack-ironic | 08:38 | |
| *** uschreiber_ has quit IRC | 08:39 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for iscsi_ilo driver https://review.openstack.org/154814 | 08:40 |
| *** uschreiber_ has joined #openstack-ironic | 08:41 | |
| *** uschreiber_ has quit IRC | 08:42 | |
| *** uschreiber_ has joined #openstack-ironic | 08:42 | |
| *** uschreiber_ has quit IRC | 08:42 | |
| *** rsalevsky has joined #openstack-ironic | 08:45 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for agent_ilo driver https://review.openstack.org/154816 | 08:47 |
| *** Nisha has joined #openstack-ironic | 08:52 | |
| Nisha | dtantsur, hi | 08:53 |
| GheRivero | morning | 08:54 |
| Nisha | GheRivero, morning | 08:54 |
| openstackgerrit | Anusha Ramineni proposed openstack/ironic: Support agent_ilo driver to perform cleaning https://review.openstack.org/164646 | 08:57 |
| *** hyakuhei has quit IRC | 09:00 | |
| *** lintan has quit IRC | 09:02 | |
| *** lucasagomes has joined #openstack-ironic | 09:12 | |
| *** andreykurilin_ has joined #openstack-ironic | 09:15 | |
| *** hyakuhei has joined #openstack-ironic | 09:18 | |
| *** hyakuhei has quit IRC | 09:28 | |
| *** MattMan has joined #openstack-ironic | 09:30 | |
| *** tiagogomes_ has joined #openstack-ironic | 09:32 | |
| *** dmellado has quit IRC | 09:38 | |
| *** saripurigopi has quit IRC | 09:45 | |
| *** alexpilotti has joined #openstack-ironic | 09:47 | |
| *** penick has quit IRC | 09:49 | |
| *** yog__ has quit IRC | 09:49 | |
| *** yog__ has joined #openstack-ironic | 09:55 | |
| *** yog__ has quit IRC | 09:59 | |
| *** Nisha has quit IRC | 09:59 | |
| *** naohirot has quit IRC | 10:00 | |
| *** dtantsur is now known as dtantsur|bbl | 10:08 | |
| openstackgerrit | Anusha Ramineni proposed openstack/ironic: Support agent_ilo driver to perform cleaning https://review.openstack.org/164646 | 10:20 |
| *** stendulker has joined #openstack-ironic | 10:26 | |
| *** chenglch has quit IRC | 10:32 | |
| *** andreykurilin_ has quit IRC | 10:34 | |
| *** pelix has joined #openstack-ironic | 10:41 | |
| *** ifarkas has quit IRC | 10:44 | |
| *** ifarkas has joined #openstack-ironic | 10:47 | |
| *** ramineni has quit IRC | 10:54 | |
| *** penick has joined #openstack-ironic | 10:58 | |
| *** athomas has quit IRC | 10:58 | |
| rameshg87 | lucasagomes: looks like target for this https://blueprints.launchpad.net/ironic/+spec/cisco-ucs-pxe-driver has been set to liberty | 10:59 |
| rameshg87 | lucasagomes: should we just -2 https://review.openstack.org/#/c/159734/ till kilo is complete ? | 10:59 |
| *** Nisha has joined #openstack-ironic | 11:01 | |
| Nisha | dtantsur|bbl, lucasagomes | 11:01 |
| Nisha | hi | 11:01 |
| lucasagomes | rameshg87, hi, yes if the work has been postponed to L | 11:01 |
| lucasagomes | Nisha, hi there | 11:01 |
| Nisha | lucasagomes, hi | 11:01 |
| Nisha | require reviews on https://review.openstack.org/151596, https://review.openstack.org/155900 and https://review.openstack.org/163572 | 11:02 |
| *** dtantsur|bbl is now known as dtantsur | 11:02 | |
| lucasagomes | I will take a look soon | 11:02 |
| Nisha | dtantsur, ^^^ | 11:02 |
| Nisha | lucasagomes, thanks today is already 18th | 11:03 |
| dtantsur | Nisha, we do see all the reviews on our dashboards/emails/launchpad. if we don't review, that doesn't mean we're evil or hate you, that only means that we are busy with other stuff | 11:03 |
| dtantsur | if you review is on priority list, it will be reviewed | 11:03 |
| Nisha | dtantsur, yeah i know | 11:04 |
| mrda | Hi lucasagomes | 11:04 |
| mrda | Want to talk about logical name? | 11:04 |
| mrda | :) | 11:04 |
| lucasagomes | mrda, hey michael | 11:04 |
| lucasagomes | yes | 11:05 |
| lucasagomes | I might be missing something there | 11:05 |
| lucasagomes | not sure | 11:05 |
| Nisha | but still asking as if there are any comments i can address them now.... | 11:05 |
| lucasagomes | mrda, if logical names aren't supported, we expect NotAcceptable or InvalidUuidOrName ? | 11:05 |
| mrda | Correct | 11:05 |
| mrda | But, we need the dual conditional not allow_node_logical_names()) and utils.is_hostname_safe(node_ident) before raising Not Acceptable | 11:06 |
| mrda | otherwise, anyhting that doesn't pass the UUID test, will result in a Not Acceptable | 11:07 |
| lucasagomes | mrda, it won't, unless logical names are not supported | 11:08 |
| mrda | I agree that if could be rewritten slightly differently so as to check is_hostname_safe once, but I tried that and other review comments wanted it made clearer | 11:08 |
| mrda | ok | 11:08 |
| rameshg87 | mrda: one thing i noted | 11:08 |
| lucasagomes | mrda, btw all ur tests passes if that bit is removed | 11:08 |
| rameshg87 | mrda: currently we don't raise NotAcceptable for GET calls | 11:08 |
| lucasagomes | mrda, http://paste.openstack.org/show/193081/ | 11:08 |
| rameshg87 | mrda: but this patch starts throwing NotAcceptable for all calls (including GET) | 11:09 |
| mrda | lucasagomes: maybe my tests are inadequate :P | 11:09 |
| * mrda looks at the pastebin | 11:09 | |
| lucasagomes | mrda, yeah | 11:09 |
| dtantsur | oh these microversions... I fully regret we started to go that deep with them... | 11:10 |
| * mrda tries to type but is too tired | 11:10 | |
| lucasagomes | mrda, so the concern is that if it's not an UUID it will raise NotAcceptable? | 11:11 |
| *** stendulker has quit IRC | 11:12 | |
| mrda | lucasagomes: Have you seen https://etherpad.openstack.org/p/ironic-microversion-handling ? | 11:12 |
| lucasagomes | mrda, checking | 11:13 |
| mrda | I'm trying to capture the different cases here | 11:13 |
| mrda | Basically, the concern I have with what you're suggesting is that if logical names aren't supported, anything that isn't a UUID will result in a HTTP 406 | 11:13 |
| *** athomas has joined #openstack-ironic | 11:14 | |
| rameshg87 | mrda: i am still not sure if last statement is completely true | 11:14 |
| rameshg87 | mrda: "We need to decide If behaviour 1B is what we want going forward. This was what was introduced in review 141737, but we need to ensure this is what we want to do going forward." | 11:14 |
| rameshg87 | mrda: we haven't changed the GET calls to raise NotAcceptable in 141737 | 11:14 |
| rameshg87 | mrda: but this patch will :) | 11:15 |
| rameshg87 | mrda: so we are introducing some new behaviours as part of this refactoring - which i doubt if we really want :) | 11:16 |
| mrda | rameshg87: we have added a 406 to a GET - GET /v1/nodes/validate in that review | 11:16 |
| * rameshg87 checks | 11:16 | |
| mrda | https://review.openstack.org/#/c/141737/29/ironic/api/controllers/v1/node.py#L860 | 11:17 |
| lucasagomes | mrda, right, so the post logical names breaks the pre because it won't raise 406 anymore. Perhaps we shouldn't introduce 406 anywhere? | 11:17 |
| *** romcheg has quit IRC | 11:17 | |
| lucasagomes | what about | 11:17 |
| lucasagomes | 1) if it's UUID: call get_by_uuid() | 11:17 |
| mrda | that's whay I wrote up https://etherpad.openstack.org/p/ironic-microversion-handling so we could discuss :) | 11:17 |
| rameshg87 | mrda: was it line 860 of node.py ? | 11:18 |
| lucasagomes | 2) if logical name isn't supported: raise Invalid (Not InvalidUuidOrName, because that would tell the user "logical name" when it's not even supported) | 11:18 |
| rameshg87 | mrda: wondering if url you gave me is wrong | 11:18 |
| mrda | oh, 867 | 11:18 |
| lucasagomes | 3) if is hosname like: call get_by_name() | 11:18 |
| mrda | sorry 923 | 11:18 |
| lucasagomes | 4) raise InvalidNameOrUuid (because now we know that logical names are suppoted but the id is not uuid-like nor hostname-safe) | 11:18 |
| rameshg87 | mrda but that's validate no | 11:19 |
| mrda | rameshg87: validate is a GET :) | 11:19 |
| rameshg87 | mrda: what about GET /v1/nodes , have introduced a 406 for that in 141737 | 11:19 |
| rameshg87 | mrda: oh you meant like that | 11:19 |
| rameshg87 | mrda: i was meaning GET = GET /v1/nodes | 11:20 |
| *** romcheg has joined #openstack-ironic | 11:20 | |
| mrda | lucasagomes: I would say that your point 2) should be raise InvalidUuid, because that was what Juno had | 11:20 |
| rameshg87 | mrda: okay so your point was it already being done on one GET call, why not on all GET calls ? | 11:20 |
| lucasagomes | mrda, right, could be | 11:20 |
| mrda | lucasagomes: if we're going to go back to this, we might as well gain some benefit of being backward compat | 11:20 |
| lucasagomes | mrda, that yes, will keep the compatibility with Juno | 11:20 |
| lucasagomes | and we don't introduce any new exception that won't be raised in the future | 11:20 |
| lucasagomes | mrda, yes | 11:21 |
| mrda | so from an operator perspective, there's plusses and minuses | 11:21 |
| mrda | if an operator tries and use logical name, but gets the microversion wrong, they'll be like InvalidUuid? WTF? | 11:21 |
| mrda | at least how it is proposed today they get NotAcceptable | 11:22 |
| mrda | But of course the other side is they get the same behaviour if they specifty a previous microversion | 11:22 |
| mrda | pros and cons | 11:22 |
| lucasagomes | we want to mimic what was before right? In Juno if someone tries to use logical name what does it get InvalidUuid | 11:23 |
| lucasagomes | that's exactly what we are doing | 11:23 |
| lucasagomes | if we return 406 on the new version we are breaking backward compat | 11:23 |
| mrda | I'm not sure, rameshg87, why we should differentiate between a 406 being raised for validate, but a 400 for GET /nodes | 11:23 |
| rameshg87 | mrda: we should go back 141737 and check if there was a reason | 11:23 |
| mrda | lucasagomes: so playing the devisl advocate we should pull out the 406s completely from node.py? | 11:24 |
| rameshg87 | mrda: but yeah, i agree with you | 11:24 |
| rameshg87 | mrda: it should be consistent | 11:24 |
| mrda | hence why the etherpad discussing the options | 11:24 |
| mrda | I don't want this to be half baked | 11:24 |
| lucasagomes | mrda, depends on the situations, I don't have all in mind now | 11:24 |
| lucasagomes | but e.g, if an endpoint doesn't exist on the old API | 11:24 |
| mrda | we should decide as a team what we expect the behaviour to be, and do it consistently across the board | 11:24 |
| lucasagomes | and we add backward compat layer there, before it would return 404 and in the new one we would return 406? | 11:25 |
| mrda | depends what you mean by endpoint lucasagomes | 11:25 |
| lucasagomes | that's bad, we should continue to reutrn 404 | 11:25 |
| mrda | is /nodes/UUID and /nodes/name different end points in your mind? | 11:25 |
| lucasagomes | mrda, like a new endpoint in the api, e.g /nodes/<uuid>/maintenance | 11:25 |
| lucasagomes | didn't exist before Kilo | 11:25 |
| lucasagomes | mrda, no they are not | 11:25 |
| mrda | then there are no new endpoints in this argument | 11:26 |
| lucasagomes | just different parameters | 11:26 |
| lucasagomes | /nodes/<uuid>/maintenance < didn't exist before | 11:26 |
| lucasagomes | before Kilo | 11:26 |
| lucasagomes | /maintenance is new endpoint | 11:26 |
| mrda | ok, that one then :) | 11:26 |
| mrda | so all the 406s introduced in 141737 should not be there | 11:26 |
| mrda | according to this line of thinking | 11:27 |
| lucasagomes | mrda, https://review.openstack.org/#/c/164369/4/ironic/api/controllers/v1/node.py | 11:27 |
| lucasagomes | this is returning 404, which is correct | 11:27 |
| lucasagomes | that's a backward compat layer | 11:27 |
| *** pas-ha has quit IRC | 11:27 | |
| lucasagomes | mrda, the 406s may be valid in some cases, I gotta take a look at the current ones | 11:27 |
| * rameshg87 goes home | 11:27 | |
| rameshg87 | will check this later :) | 11:27 |
| *** rameshg87 has quit IRC | 11:27 | |
| mrda | lucasagomes: sure, I've seen this review | 11:28 |
| mrda | But it shounds like you're advocating the unwinding of 406s in 141737. Is that right? | 11:28 |
| lucasagomes | mrda, yes | 11:29 |
| lucasagomes | mrda, I don't think it's needed | 11:29 |
| mrda | validate, post and patch | 11:29 |
| mrda | That's all | 11:29 |
| mrda | BTW, I'm not opposed to this thought. I just want consensus | 11:30 |
| lucasagomes | 141737 ? | 11:30 |
| lucasagomes | mrda, sorry backing my answer | 11:30 |
| lucasagomes | for 141737 the answer is I don't know | 11:30 |
| lucasagomes | cause I don't know all the uses for 406 yet | 11:31 |
| lucasagomes | I believe that it's valid for some cases | 11:31 |
| lucasagomes | If it doesn't break backward compat, it can be a valid case | 11:31 |
| lucasagomes | mrda, but specifically to the logical names patch, 406 is not the expected return value for some code paths. And is not backward compat | 11:32 |
| lucasagomes | plus will be removed later | 11:33 |
| lucasagomes | (according to that etherpad, pos-logical-names) | 11:33 |
| mrda | I don't know how to proceed on 163730 | 11:34 |
| mrda | I think that if you want to -1 https://review.openstack.org/#/c/163730/ you need to update https://etherpad.openstack.org/p/ironic-microversion-handling with an alternative proposal. | 11:35 |
| lucasagomes | mrda, hmm yeah maybe we should talk more about it. I will leave another comment re NotAcceptable | 11:35 |
| mrda | and that way other stakeholders in ironic can decide upon the merit, and make a choice | 11:36 |
| lucasagomes | mrda, right... yeah I will try to ping devananda and others to take a look | 11:38 |
| lucasagomes | mrda, but, if I had to vote what to do, I would vote for being totally backwards compat | 11:38 |
| lucasagomes | because I believe that's the reason why we introduced micro versioning | 11:38 |
| mrda | I think we kind agreed to accept this as is for now, and iterate over it in the next week while we fix up python-ironicclient | 11:39 |
| *** persia has quit IRC | 11:39 | |
| mrda | but if you feel strongly about it I would recommend arguing it in the etherpad so devananda et al can comment when they are online | 11:39 |
| *** pas-ha has joined #openstack-ironic | 11:40 | |
| *** persia_ has joined #openstack-ironic | 11:40 | |
| *** persia_ has quit IRC | 11:40 | |
| *** persia_ has joined #openstack-ironic | 11:40 | |
| lucasagomes | mrda, right, ok the folks from US might be waking up soonish | 11:40 |
| lucasagomes | so we can discuss it, and if we decide to go with ur way I'm happy to change my vote | 11:41 |
| mrda | I going to head off of bed now. Thanks for the discussion, lucasgomes. I look forward to reading what you add :) I always appreciate what you have to say :) | 11:41 |
| mrda | I honestly don't mind, I just want the code to be consistent :) | 11:41 |
| lucasagomes | mrda, oh yeah it's late there. Have a good night michael | 11:42 |
| lucasagomes | thanks much | 11:42 |
| mrda | have a nice day lucasagomes! | 11:42 |
| lucasagomes | sorry for the turbulence on that patch, I will make sure I add the comments on the review/etherpad | 11:42 |
| *** mrda is now known as mrda-away | 11:42 | |
| *** persia_ has quit IRC | 11:44 | |
| *** achanda has joined #openstack-ironic | 11:45 | |
| *** persia has joined #openstack-ironic | 11:46 | |
| *** achanda has quit IRC | 11:49 | |
| *** Nisha has quit IRC | 11:51 | |
| *** Nisha_away has joined #openstack-ironic | 11:51 | |
| *** dprince has joined #openstack-ironic | 11:56 | |
| *** vdrok_afk is now known as vdrok | 11:59 | |
| vdrok | morning ironic | 12:00 |
| *** david-lyle_afk is now known as david-lyle | 12:00 | |
| *** Nisha_away has quit IRC | 12:01 | |
| *** romcheg has quit IRC | 12:03 | |
| *** pradipta has quit IRC | 12:04 | |
| *** ramineni has joined #openstack-ironic | 12:08 | |
| *** stendulker has joined #openstack-ironic | 12:11 | |
| ramineni | dtantsur: hi | 12:12 |
| dtantsur | o/ | 12:12 |
| ramineni | dtantsur: Regarding your comments on https://review.openstack.org/#/c/157715/17/ | 12:13 |
| ramineni | dtantsur: this is being introduced in a parent blueprint patch: https://review.openstack.org/#/c/161453/12/ironic/common/exception.py | 12:13 |
| dtantsur | yes | 12:13 |
| ramineni | dtantsur: i thought about that , but nothing from my patch is dependent on that except the exception part | 12:14 |
| *** romcheg has joined #openstack-ironic | 12:14 | |
| ramineni | dtantsur: thats y didnt rebase on top of it | 12:14 |
| dtantsur | it's not a compelling reason to introduce exception twice ;) | 12:14 |
| ramineni | dtantsur: the parent was about inband cleaning ..loading ramdisk and all , but this patch doesnt need anything of that sort | 12:15 |
| ramineni | dtantsur: :) i agree , patch getting merged first to be rebased | 12:15 |
| dtantsur | ramineni, you can suggest the author to move introducing exception in a separate patch... but I doubt he will agree :) I'm trying to avoid unnecessary conflict for one of you | 12:16 |
| ramineni | dtantsur: ya, right | 12:16 |
| ramineni | dtantsur: ok, no problem. ill rebase | 12:17 |
| ramineni | dtantsur: and regarding error part | 12:17 |
| *** bradjones has joined #openstack-ironic | 12:18 | |
| ramineni | dtantsur: we are not failing , but continuing with other tests , so warning might be appropriate? | 12:18 |
| *** derekh has joined #openstack-ironic | 12:19 | |
| dtantsur | ramineni, what are you trying to tell the operator by warning, that's the question. If you think skipping this step is completely ok, then it can be INFO | 12:19 |
| dtantsur | ramineni, but if you think it's an error, then it should be ERROR | 12:19 |
| dtantsur | imaging person not wanting to ever changing passwords. he/she will receive these warnings every cleaning | 12:20 |
| dtantsur | and warning is something that should get attention. that's how I reason | 12:20 |
| dtantsur | now sorry, quick food :) | 12:20 |
| ramineni | dtantsur: hmm, ok | 12:21 |
| ramineni | dtantsur: :) , will ping you later | 12:23 |
| *** dmellado has joined #openstack-ironic | 12:26 | |
| openstackgerrit | PSPradhan proposed stackforge/proliantutils: Update RIS library https://review.openstack.org/163290 | 12:35 |
| *** stendulker has quit IRC | 12:37 | |
| *** lucasagomes is now known as lucas-hungry | 12:39 | |
| *** jistr is now known as jistr|biab | 12:39 | |
| *** rloo has joined #openstack-ironic | 12:44 | |
| *** rameshg87 has joined #openstack-ironic | 12:47 | |
| *** dlpartain has joined #openstack-ironic | 12:48 | |
| openstackgerrit | Ghe Rivero proposed openstack/ironic: Sync from oslo.incubator https://review.openstack.org/162505 | 13:05 |
| * rameshg87 rameshg87-away | 13:08 | |
| *** kkoski has joined #openstack-ironic | 13:08 | |
| *** rameshg87 is now known as rameshg87-away | 13:08 | |
| ramineni | dtantsur: replied to your comments on https://review.openstack.org/#/c/157715/ , let me know if you are fine | 13:10 |
| openstackgerrit | PSPradhan proposed stackforge/proliantutils: Update RIS library https://review.openstack.org/163290 | 13:16 |
| *** absubram has joined #openstack-ironic | 13:19 | |
| Shrews | devananda: fyi, the node name stuff from https://review.openstack.org/158520 isn't working for me. Not sure if this is a client or server problem. | 13:21 |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: [WIP] Add DevStack plugin for ironic-discoverd https://review.openstack.org/164782 | 13:22 |
| Shrews | or even a PEBKAC problem | 13:22 |
| Shrews | also, morning ironicers | 13:23 |
| kkoski | Good morning Shrews | 13:24 |
| *** stendulker has joined #openstack-ironic | 13:24 | |
| Shrews | morning kkoski | 13:24 |
| *** dlpartain has left #openstack-ironic | 13:26 | |
| rloo | morning Shrews and everyone else | 13:26 |
| rloo | Shrews: I replied in the patch; it works, you need to specify a microversion | 13:26 |
| rloo | Shrews: I think. i haven't looked at the patch yet. | 13:27 |
| Shrews | rloo: ah! it *was* a pebkac error. and morning | 13:27 |
| rloo | Shrews: well, i think it is interesting to see what happens when someone tries something and they don't get the expected result. Was the error not very informative? | 13:28 |
| *** ChuckC has joined #openstack-ironic | 13:28 | |
| * Shrews HATES having to specify 'latest' for version | 13:28 | |
| Shrews | we really need to change that | 13:28 |
| Shrews | rloo: i just forgot that we do things wrong | 13:28 |
| rloo | Shrews: you should be able to set some environment variable to 'latest'. i haven't tried it yet but it should work if i recall the state of the client code. | 13:29 |
| Shrews | yeah | 13:30 |
| rloo | Shrews: IRONIC_API_VERSION | 13:30 |
| rloo | Shrews: please test it ;) | 13:30 |
| openstackgerrit | Anusha Ramineni proposed openstack/ironic: Add Cleaning Operations for iLO drivers https://review.openstack.org/157715 | 13:30 |
| Shrews | rloo: yeah, that works | 13:31 |
| rloo | Shrews: cool. something works :D | 13:31 |
| *** alexpilotti has quit IRC | 13:35 | |
| *** stendulker has quit IRC | 13:36 | |
| *** lucas-hungry is now known as lucasagomes | 13:37 | |
| lucasagomes | morning Shrews rloo kkoski ramineni | 13:37 |
| ramineni | lucasagomes: morning :) | 13:38 |
| lucasagomes | rloo, Shrews me and mrda-away talked about https://review.openstack.org/#/c/163730/ today, when you guys have a time, mind taking a look at the comments? | 13:38 |
| Shrews | lucasagomes: will do | 13:38 |
| rloo | hi lucasagomes. will look in a few minutes. | 13:39 |
| lucasagomes | rloo, Shrews no rush, thanks | 13:39 |
| *** jistr|biab is now known as jistr | 13:39 | |
| dtantsur | Shrews, rloo, morning | 13:43 |
| Shrews | hi dtantsur | 13:43 |
| rloo | hi dtantsur :) | 13:43 |
| devananda | g'morning all | 13:48 |
| dtantsur | devananda, o/ | 13:48 |
| devananda | Shrews: "we do things wrong" -- please elaborate | 13:49 |
| devananda | dtantsur: \o | 13:49 |
| Shrews | devananda: this was discussed previously. i expect a client to always use the latest API that the server supports | 13:49 |
| Shrews | that fact that we don't, IMO, is broken behaviour | 13:50 |
| devananda | dtantsur: I put a much longer explanation in my patch lowering the version to 1.0 -- any further thoughts on it? | 13:50 |
| *** kkoski has quit IRC | 13:50 | |
| devananda | Shrews: hum. latest that the server supports, or latest common version that both client and server support? | 13:50 |
| * dtantsur is looking | 13:50 | |
| Shrews | devananda: the latter | 13:52 |
| dtantsur | devananda, I can't find answer to question "why gate things that do not break backward compatibility for any sane definition of backward compatibility" :) am I missing something? | 13:52 |
| rloo | hi lucasagomes. wrt 163730, it seems like "all" it was meant to do was to support logical names. if people are disagreeing with the behaviour that was there already (raising 400 vs 406) shouldn't that be addressed in a separate patch/bug? | 13:52 |
| devananda | Shrews: ok. ten you and I agree. the missing piece is that our client need to default to sending a version string, and right now, does not | 13:52 |
| dtantsur | devananda, in other works, I think we're taking introducing new API versions to easily. What if I say that we need a separate Tempest coverage for each micro version? | 13:53 |
| dtantsur | * words obviously | 13:53 |
| devananda | dtantsur: ah! no - you're not. that question has been on my mind as well.... however... | 13:53 |
| devananda | dtantsur: (you're not missing something) | 13:53 |
| jroll | morning devananda dtantsur Shrews rloo ramineni and anyone else that may be above the fold :) | 13:54 |
| dtantsur | o/ | 13:54 |
| *** kkoski has joined #openstack-ironic | 13:55 | |
| *** rameshg87-away is now known as rameshg87 | 13:55 | |
| ramineni | morning jroll | 13:55 |
| devananda | dtantsur: let's say I write a client that expects a particular format of response (say, containing 10 fields or something) | 13:55 |
| dtantsur | I hope you don't | 13:55 |
| rloo | morning jroll, welcome to the party! | 13:55 |
| dtantsur | that does not fit in my definition of sanity, sorry | 13:55 |
| devananda | dtantsur: then let's say we add a new field to the result of GET /v1/nodes/NNNN -- that's 'backwrds compatible" because it didn't change the existing API ... | 13:55 |
| devananda | dtantsur: but yea, it just broke that client | 13:55 |
| devananda | dtantsur: or to put it differently, the client got a result which is not defined by the API at the time the client was written | 13:56 |
| devananda | dtantsur: so by not guarding new fields in the result of GET /v1/nodes/NNN we are implicitly requiring all clients to handle the unknown -- rather than explicitly telling clients "here is the API structure you can count on" | 13:57 |
| dtantsur | devananda, this way we can go as far as "what if clients rely on Ironic ignoring X-OpenStack-Ironic-Version header" :D and seriously we need to switch to SOAP then I not use technologies designed to be extensible | 13:57 |
| Shrews | rloo: lucasagomes: So, type validation is done pretty early. This is an interesting question, but I sort of agree with rloo that this might be a change for a different patch. It could be significant. | 13:57 |
| *** zz_jgrimm is now known as jgrimm | 13:57 | |
| dtantsur | devananda, even modern binary formats like Protocol Buffers account for unknown optional fields | 13:57 |
| dtantsur | if we want a fixed schema, we need a protocol supporting fixed schema | 13:58 |
| Shrews | ooh, protocol buffers.... reminds me of my drizzle days :) | 13:58 |
| devananda | Shrews: I had the same thought :) | 13:58 |
| dtantsur | do we have examples outside of OpenStack doing this kind of stuff with their REST API? | 13:59 |
| Shrews | hmm, maybe brian was right that using JSON instead of PB was the wrong way :) | 13:59 |
| devananda | Shrews: PB for a public api? hrm | 13:59 |
| dtantsur | I tried on my first job. I'd better use JSON... | 14:00 |
| dtantsur | (but anything was better than SOAP :) | 14:00 |
| Shrews | dtantsur++ | 14:00 |
| Shrews | i always felt dirty when doing anything with SOAP | 14:00 |
| devananda | dtantsur: I can see benefits to both sides of this discussion, honestly | 14:01 |
| dtantsur | getting back to our discussion: I feel like if we create protocol with fixed schema, I think we're taking it too lightly... | 14:01 |
| dtantsur | I really think that we should have tempest coverage for every micro version | 14:01 |
| dtantsur | (and that is one of the reasons to have only 2 microversions: 1.0 = Juno, 1.1 = Kilo) | 14:02 |
| *** hyakuhei has joined #openstack-ironic | 14:02 | |
| devananda | dtantsur: and while trying to balance "allow server API to change gradually" with "give users (and their client tooling) predictable experience from the API", I think either approach is going to have some pain and some win | 14:02 |
| rloo | devananda: do you know what nova is doing? if OpenStack is using microversions, it seems like it would be a good idea to implement it consistently in the projects | 14:02 |
| *** viktors has joined #openstack-ironic | 14:02 | |
| *** hyakuhei has quit IRC | 14:02 | |
| devananda | where either == "use microversions only for things that actually break, like s/NOSTATE/AVAILABLE/" OR "use microversions for any change" | 14:02 |
| devananda | rloo: ++ | 14:02 |
| devananda | rloo: I'll double check, but i'm fairly sure nova wants to have lots of little microversions (not just one per release) | 14:03 |
| devananda | their spec clearly states that | 14:03 |
| dtantsur | devananda, so my next question is: with this microversions we've broken inter-commit compatibility several times already. Why not squash into 2 versions? | 14:03 |
| *** alexpilotti has joined #openstack-ironic | 14:03 | |
| rloo | devananda: yes, that's my understanding. And it think it includes hiding new fields or whatever from responses | 14:04 |
| devananda | because anyone doing CD from trunk will be broken if there are two different "noticieable" changes within the same microversion | 14:04 |
| devananda | rloo: yup | 14:04 |
| dtantsur | devananda, we can force-bump on any breaking change | 14:04 |
| devananda | dtantsur: the real question is: what does "breaking" mean? | 14:04 |
| devananda | dtantsur: or rather, who is actually broken by it? | 14:05 |
| lucasagomes | Shrews, rloo will read in a second I was in a call | 14:05 |
| dtantsur | devananda, I define it like "correct and sane code written before the change will stop working" | 14:05 |
| devananda | the vast majority of our users, as far as I'm aware, are not continuously deploying ironic. they're consuming packages | 14:05 |
| Shrews | lucasagomes: still reading the comments, too :) | 14:05 |
| jroll | random question that probably will get things thrown at me: what if we did super-proper semver? x.y.z, where x is major version, y is breaking changes, z is things like adding an endpoint. and we guard for x and y. | 14:05 |
| rloo | I do agree with dtantsur about what level of changes consists of one microversion change. | 14:05 |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic: iscsi_ilo driver do not validate boot_option https://review.openstack.org/164414 | 14:05 |
| rameshg87 | dtantsur: need your +2 again on https://review.openstack.org/#/c/164414/ | 14:06 |
| dtantsur | jroll, then what's the point of 'z' component? | 14:06 |
| rameshg87 | dtantsur: the driver name was wrong | 14:06 |
| rameshg87 | dtantsur: it was written ilo_iscsi (instead of iscsi_ilo) | 14:06 |
| jroll | dtantsur: to indicate that 'z' has a new feature or whatever. just throwing the question out there. full disclosure: not enough coffee yet. | 14:06 |
| dtantsur | rameshg87, done | 14:06 |
| rameshg87 | dtantsur: thanks :) | 14:07 |
| lucasagomes | dtantsur, ++ | 14:09 |
| dtantsur | devananda, my concern with too many versions is that we can easily break, say, 1.3 and nobody will notice, because: 1. we're not testing it; 2. nobody is actually using it. | 14:09 |
| dtantsur | which will mean that we have versions of the sake of versions | 14:09 |
| *** stendulker has joined #openstack-ironic | 14:09 | |
| devananda | dtantsur: oh. you're absolteuly right -- today, we're not testing these well enough | 14:09 |
| dtantsur | s/enough/at all/ once we introduce the next one, no? | 14:10 |
| devananda | dtantsur: if we all collectively actually followed sane test-before-you-land development practices, then we would require a proposal for tempest API tests of any API change befor ethe API code landed | 14:10 |
| devananda | dtantsur: and we'd require a client patch that implemented support for said API before landing the server change too | 14:10 |
| devananda | dtantsur: and we'd test the proposed client against stable branch. and we'd also test stable branch of client against the proposed server API | 14:11 |
| devananda | all before we land the API change | 14:11 |
| devananda | and i would be DELIGHTED if we did all that | 14:11 |
| lucasagomes | Shrews, rloo right, but how we can do in a different patch if by supporting logical name we have to verify whether it's supported by the API or not (microversioning) | 14:11 |
| dtantsur | hmm, I mean a bit different thing | 14:11 |
| lucasagomes | maybe there's 2 conversations going on at the same time, let's talk about it when the first one is finished | 14:11 |
| Shrews | lucasagomes: ok. i left comments in the review, too | 14:12 |
| dtantsur | devananda, I mean that we can at some point break the whole 1.3 (for example) API completely (half of endpoints raise error 500), and no one will notice | 14:12 |
| NobodyCam | good morning Ironic | 14:12 |
| rloo | lucasagomes: yes, I agree, if it is a bug, it should be addressed in a separate patch. I just commented in the patch about that. | 14:12 |
| devananda | dtantsur: you are correct. I think it's crazypants. | 14:12 |
| *** BadCub_Sleeping is now known as BadCub | 14:12 | |
| devananda | dtantsur: which is why I think we need tempest API tests passing "latest" versio string | 14:12 |
| devananda | dtantsur: so that we prevent that sort of crazy thing where v1.9 removes /v1/nodes/ endpoint entirely | 14:13 |
| lucasagomes | Shrews, rloo thanks | 14:13 |
| *** mtanino has joined #openstack-ironic | 14:13 | |
| BadCub | Morning everyone | 14:13 |
| devananda | dtantsur: right now, we are completely NOT testing the current API at all. None of our test suite, as far as I know, tests ANYTHING BESIDES v1.1 | 14:13 |
| dtantsur | devananda, it does not prevent us from breaking some previous version later one... e.g. we implement incorrect version check that will cause error 500 for some old version | 14:13 |
| NobodyCam | morning BadCub | 14:13 |
| TheJulia | good morning BadCub | 14:13 |
| dtantsur | NobodyCam, BadCub, TheJulia, morning | 14:14 |
| *** mgoddard has quit IRC | 14:14 | |
| NobodyCam | morning dtantsur | 14:14 |
| rloo | lucasagomes: feel free to open a bug about it. I haven't been paying attention. I know there was some discussion about it in the weekly meeting but not sure what was decided if anything. | 14:14 |
| NobodyCam | morning TheJulia | 14:14 |
| *** mgoddard has joined #openstack-ironic | 14:14 | |
| *** alexpilotti has quit IRC | 14:14 | |
| devananda | dtantsur: we need the client to pass the latest version string it knows about by default, so that we can properly test forwards-and-backwards compat between stable branches | 14:14 |
| TheJulia | Good morning * (Not enough coffee reaching brain yet) | 14:14 |
| devananda | dtantsur: and we need tempest to pass "latest" so we can test the proposed API changes | 14:14 |
| lucasagomes | rloo, right. Oh, I will check the logs from the last meeting | 14:14 |
| rloo | lucasagomes: I think the meeting notes said that they were taking it to irc/etherpad or something ;) | 14:15 |
| lucasagomes | rloo, https://etherpad.openstack.org/p/ironic-microversion-handling there's this etherpad | 14:15 |
| BadCub | Morning TheJulia , NobodyCam, dtantsur , | 14:15 |
| rloo | hey, are we still trying to get cleaning stuff landed for k-3? | 14:15 |
| *** alexpilotti has joined #openstack-ironic | 14:15 | |
| devananda | dtantsur: and when we're doing both of those things (which we're not today) then I will still feel the same way that I do now about microversions. (rather than say "microversions are only for big changes, we're not testing them" I am saying "we are doing a bad job of testing right now, but microversions are still good for small changes") | 14:15 |
| lucasagomes | rloo, mrda-away pointed me to it, but I find the strategy there odd. The pos-logical name breaks the pre-logical name | 14:16 |
| dtantsur | devananda, hmm, still some misunderstanding... Ok, imagine latest is 1.8. We're working on 1.9 and introduce change (e.g. in version check code) that breaks version 1.3 (happened in the middle of Kilo, so ~ nobody is using it). How to prevent it? | 14:16 |
| devananda | rloo: if we can | 14:16 |
| rloo | lucasagomes: can we change/decide what to do after k-3 and before k release? | 14:16 |
| devananda | dtantsur: oooh. gotcha | 14:16 |
| devananda | dtantsur: we might accidentally break an old mid-release version later on -- that's what you mean | 14:17 |
| dtantsur | yep | 14:17 |
| devananda | dtantsur: yah. fair point | 14:17 |
| rloo | lucasagomes: i think i should focus on cleaning patches today... | 14:17 |
| devananda | dtantsur: we could test every version... | 14:17 |
| lucasagomes | rloo, fair enough. Yeah we can do that. /me forgot when k3 will be released | 14:17 |
| * lucasagomes checks | 14:17 | |
| dtantsur | devananda, we probably should :) and to me it's the reason to have less of them | 14:17 |
| lucasagomes | tomorrow | 14:18 |
| rloo | lucasagomes: yeah, or whenever devananda pulls the switch. | 14:18 |
| devananda | dtantsur: but I think it has an exponentially increasing cost for decreasing benefit (risk of harm is much lower, risk of problem is lower, cost of preventing it is much higher than fixing it) | 14:18 |
| devananda | dtantsur: but yah, I see your point | 14:18 |
| devananda | rloo, lucasagomes: k3 will be tagged some time this week, pretty much based on when we / I think it's ready | 14:19 |
| devananda | if there are things that anyone feels absolutely must be landed, they should be being tracked on launchpad | 14:19 |
| BadCub | lucasagomes: could you take a look at https://review.openstack.org/#/c/151596/ It needs one more +2/+A | 14:20 |
| lucasagomes | devananda, fair enuff. We are discussing a patch about logical names. I think it's close to land but there's some misunderstanding there about the return code. If you have some time please take a look at the comments https://review.openstack.org/#/c/163730/ | 14:20 |
| lucasagomes | BadCub, morning, sure thing! Will look now | 14:20 |
| BadCub | awesomeness! lucasagomes :-) | 14:21 |
| devananda | mordred: you dont want to read any of the above discussion about microversions or protocol buffers. you really don't :) | 14:21 |
| devananda | lucasagomes: yea, return code discussion is on an etherpad too | 14:21 |
| devananda | lucasagomes: https://etherpad.openstack.org/p/ironic-microversion-handling | 14:21 |
| lucasagomes | devananda, yeah mrda-away pointed me to that, but I think it's wrong in the etherpad | 14:21 |
| lucasagomes | I don't think we need the NotAcceptable exception there | 14:22 |
| lucasagomes | it breaks backward compatibility | 14:22 |
| Shrews | lucasagomes++ | 14:22 |
| lucasagomes | plus, it seems to be removed in the "Post-Logical Name microversioning" which makes it even more odd | 14:23 |
| *** lazy_prince is now known as killer_prince | 14:23 | |
| lucasagomes | to introduce something and remove later, post, basically breaks pre | 14:23 |
| * BadCub needs more coffee. BRB | 14:23 | |
| openstackgerrit | Jim Rollenhagen proposed openstack/ironic: DO NOT MERGE: verifying lack of tests. https://review.openstack.org/165430 | 14:23 |
| jroll | I never realized this. verifying. | 14:23 |
| jroll | (but I tend to think it's true) | 14:23 |
| NobodyCam | morning jroll | 14:23 |
| jroll | morning NobodyCam :) | 14:23 |
| devananda | jroll: this is why I have been arguing so strongly that we need the client to default to sending the latest version it knows about | 14:24 |
| * Shrews wants to put #-ironic on a controlled-release-delay... too much stuff gets discussed too quickly before i am properly caffienated | 14:24 | |
| devananda | jroll: right now, client sends NO version by default, meaning all our functional tests are essentially against v1.1 | 14:24 |
| mordred | devananda: what did I do? | 14:24 |
| jroll | devananda: yeah. | 14:24 |
| jroll | mordred: nothing, we just need to borrow some cats | 14:25 |
| devananda | mordred: spread (in)sanity to the world. what else? | 14:25 |
| devananda | by that I mean cats | 14:25 |
| * mordred clutches a pile of cats to his chest, starts coughing | 14:25 | |
| rloo | please no more cats. my cat brought in a mouse yesterday and I stepped on the remains of it :-( | 14:25 |
| devananda | jroll: ditto for tempest. i'm pretty sure it's not passing "latest" ... | 14:26 |
| jroll | :| | 14:26 |
| jroll | devananda: it isn't. | 14:26 |
| devananda | rloo: i'm sure your cat was very pleased | 14:26 |
| jroll | I'm fairly certain this will pass tempest | 14:26 |
| devananda | jroll: yah | 14:26 |
| mordred | devananda: I'm less angry about nova microversions after talking to sdague a bit ... it _seems_ that I may not need to know about it or pay attention perhaps | 14:26 |
| devananda | mordred: oh? | 14:26 |
| devananda | mordred: how's that? | 14:26 |
| *** rfchapman has joined #openstack-ironic | 14:26 | |
| mordred | devananda: my biggest concern was making the stupid opaque parameter that is required to be passed in to the Client constructor worse | 14:27 |
| jroll | devananda: and I want to take this change and the test results and pin it to the proverbial bulletin board | 14:27 |
| rloo | devananda: i read that we're supposed to be pleased, but yuck. we shouldn't have put in a cat door. the remains were "interesting". | 14:27 |
| mordred | right now, that I have to pass in "1" or "2" is just plain insane | 14:27 |
| *** stendulker has quit IRC | 14:27 | |
| mordred | and every single openstack project that has adopted that model, including yours, is crazy | 14:27 |
| mordred | however | 14:27 |
| rameshg87 | anyone looking into nova cleaning patch 1 https://review.openstack.org/#/c/164313/5 ? jenkins is -1 | 14:27 |
| mordred | it seems that the goal over there is to get to a place where that can be removed and the client lib can just figure it out | 14:27 |
| mordred | which is what I want | 14:27 |
| devananda | mordred: not mine. I expressly do not want users to ever have to tell their client what versio to use | 14:27 |
| jroll | rameshg87: I'll look | 14:28 |
| mordred | devananda: yes. but they have to right now, so until you fix it, you're insane too :) | 14:28 |
| rameshg87 | jroll: okay | 14:28 |
| devananda | mordred: read our latest spec -- http://git.openstack.org/cgit/openstack/ironic-specs/tree/specs/kilo/api-microversions.rst | 14:28 |
| devananda | mordred: correct. right now, we're insane in a lot of ways | 14:28 |
| devananda | mordred: I keep arguing with folks about what the client should do .... and what we should be testing ... | 14:29 |
| devananda | I believe that spec, while reading it generates insanity, actually describes a sane world. some of us must pay the price to get there, thoug h... | 14:29 |
| mordred | devananda: the chunk where you say "it should just work" gets my vote | 14:29 |
| * jroll inserts a joke about apple here | 14:30 | |
| Shrews | that should be our guiding statement on everything | 14:30 |
| * devananda begins the morning coffee making ritual | 14:30 | |
| mordred | because the '1' here: http://git.openstack.org/cgit/openstack-infra/shade/tree/shade/__init__.py#n1513 should make everyone sad | 14:30 |
| Shrews | jroll: we will soon begin building our own hardware to work with ironic | 14:30 |
| lucasagomes | hah | 14:31 |
| devananda | my new mottos: "it should just work", "get out of the way", and "do eeeeeet" | 14:31 |
| mordred | devananda: also - do you support keystone sessions yet? | 14:31 |
| mordred | devananda: because that code is still passing in a token | 14:31 |
| mordred | and we should really get it updated to pass in a Session | 14:31 |
| Shrews | mordred: https://review.openstack.org/156344 | 14:31 |
| Shrews | well, that's the server | 14:32 |
| devananda | mordred: no. someone came by and mentiond it and proposed a change, but well, we've been obsessed with this absurd kilo3 thing... | 14:32 |
| NobodyCam | morning mordred, Shrews, and lucasagomes ) | 14:32 |
| Shrews | i thought there was a client change somewhere | 14:32 |
| BadCub | morning mordred | 14:32 |
| Shrews | hey NobodyCam | 14:33 |
| NobodyCam | :) | 14:33 |
| devananda | apparently, not having coffee for an hour after waking up generates sarcasm. | 14:33 |
| mordred | in case anybody cares, here's what I think the invocation of teh client constructor _should_ look like: http://git.openstack.org/cgit/openstack-infra/shade/tree/shade/__init__.py#n503 | 14:33 |
| * devananda waits impatiently for water to boil | 14:33 | |
| NobodyCam | devananda: dont watch the pot it will necer boil | 14:33 |
| devananda | mordred: yup | 14:33 |
| NobodyCam | never even | 14:33 |
| devananda | mordred: http://git.openstack.org/cgit/openstack/ironic-specs/tree/specs/kilo/api-microversions.rst#n79 | 14:33 |
| mordred | devananda: yup | 14:34 |
| devananda | mordred: and this http://git.openstack.org/cgit/openstack/ironic-specs/tree/specs/kilo/api-microversions.rst#n197 | 14:34 |
| jroll | rameshg87: LOL | 14:34 |
| jroll | rameshg87: watch the fix | 14:34 |
| mordred | devananda: it's good to know that we all agree | 14:34 |
| NobodyCam | oh morning rameshg87 :) | 14:34 |
| devananda | :) | 14:35 |
| devananda | mordred: and this http://git.openstack.org/cgit/openstack/ironic-specs/tree/specs/kilo/api-microversions.rst#n239 --- is what we should use in the gate to test with, but a user should never need | 14:35 |
| *** kkoski has quit IRC | 14:35 | |
| devananda | in _a_ gate, I mean | 14:36 |
| mordred | devananda: yah - also - I like "report back to client" | 14:36 |
| devananda | cause I also want to test forwards-and-backwards major versions | 14:36 |
| *** bauzas has quit IRC | 14:36 | |
| rameshg87 | NobodyCam: morning | 14:36 |
| rfchapman | NobodyCam: Morning how are you doing? | 14:36 |
| mordred | devananda: error message of "dude, upgrade your ironicclient" I think is actually fair | 14:37 |
| NobodyCam | morning rfchapman :) | 14:37 |
| NobodyCam | doing good here. :) | 14:37 |
| rameshg87 | jroll: fix for what ? | 14:37 |
| mordred | as long as upgrading the ironicclient wouldnt' cause that dude to be unable to talk to her old ironic over in the corner | 14:37 |
| rameshg87 | jroll: i missed something , reading back | 14:37 |
| *** kkoski has joined #openstack-ironic | 14:37 | |
| devananda | mordred: http://git.openstack.org/cgit/openstack/ironic-specs/tree/specs/kilo/api-microversions.rst#n194 | 14:37 |
| jroll | rameshg87: I'm fixing the cleaning/nova patch now | 14:38 |
| mordred | devananda: I can't possibly be expected to read that many words this early in the morning ;) | 14:38 |
| rloo | jroll: there's a note in the etherpad that the nova patches won't be approved before liberty | 14:38 |
| jroll | rloo: that changed | 14:38 |
| rameshg87 | jroll: oh still to come out .. okay :) | 14:38 |
| jroll | rloo: or something changed last night after I left | 14:39 |
| rfchapman | NobodyCam: I'm trying to create a fedora19 image with a local qcow2 image that I have running. It's not a cloud specific image that I'm working just a kvm image. Can this be done? | 14:39 |
| rloo | jroll: oh. so I should remove the "BLOCKED UNTIL LIBERTY" thing? | 14:39 |
| jroll | rloo: yes pls | 14:39 |
| *** kkoski has quit IRC | 14:39 | |
| rloo | jroll: done. (I hope you're right!) | 14:39 |
| rfchapman | NobodyCam: I'm using disk-image-create | 14:39 |
| devananda | mordred: i haven't had coffee yet :) | 14:39 |
| devananda | mordred: in any case, i'm also glad that we agree | 14:40 |
| BadCub | devananda: how is it even possible to talk to anyone without coffee? *grumbles at his coffee mug* | 14:40 |
| NobodyCam | rfchapman: that should work... so if I understand your making a "Whole disk" image not a partition image | 14:40 |
| jroll | rameshg87: line 203 https://review.openstack.org/#/c/164313/6/nova/virt/ironic/driver.py | 14:41 |
| jroll | rameshg87: microversions win again | 14:41 |
| TheJulia | BadCub: magic makes talking w/o coffee possible | 14:42 |
| rfchapman | NobodyCam: Yes, at least I think that is what I'm doing. After reading the instructions the command line that I'm working with is DIB_LOCAL_IMAGE=/root/fedora19.qcow2 disk-image-create fedora baremetal dhcp-all-interfaces -o fedora19-image | 14:42 |
| *** killer_prince is now known as lazy_prince | 14:42 | |
| BadCub | TheJulia: Not even magic could make me coherent enough to talk w/o the blessings of coffee LOL | 14:42 |
| jroll | ^ | 14:42 |
| NobodyCam | rfchapman: for whole disk image you'll also what the VM element | 14:43 |
| * mordred hands BadCub a magical pony | 14:43 | |
| *** Nisha has joined #openstack-ironic | 14:43 | |
| rameshg87 | jroll: oh | 14:43 |
| mordred | NobodyCam: have I mentioned that that element is confusingly named | 14:43 |
| BadCub | mordred: does it have coffee? LOL | 14:43 |
| NobodyCam | mordred: +++ | 14:43 |
| jroll | lol | 14:43 |
| mordred | BadCub: nope. only magic | 14:44 |
| NobodyCam | lol BadCub | 14:44 |
| jroll | I tend to think coffee is magic, so | 14:44 |
| devananda | so yea, that port name patch? it's broken the API -- http://paste.openstack.org/show/193128/ | 14:44 |
| rfchapman | NobodyCam: By VM element do you mean the definition xml file? | 14:44 |
| BadCub | lol | 14:45 |
| jroll | devananda: why did I let y'all do this naming thing anyway :/ | 14:45 |
| devananda | jroll: I kinda hate it now, tbh | 14:45 |
| * jroll pets server-123 | 14:45 | |
| devananda | jroll: i mean, it's probably useful to some folks, but really....? | 14:45 |
| NobodyCam | rfchapman: something like : "disk-image-create fedora baremetal dhcp-all-interfaces vm -o fedora19-image" | 14:45 |
| * jroll gives db01.ironic02 a cookie | 14:45 | |
| NobodyCam | lol | 14:45 |
| NobodyCam | jroll: do I need to pick up more pet food at the store? | 14:46 |
| BadCub | *buries head deep in coffee mug while folks pet and feed servers* | 14:46 |
| jroll | NobodyCam: just toys, I can't find server-123's favorite chew port | 14:46 |
| NobodyCam | lol | 14:46 |
| devananda | jroll: ironic node-update fake replace name=db01.ironic | 14:46 |
| devananda | Node fake: Cannot change name to invalid name 'db01.ironic' (HTTP 400) | 14:46 |
| devananda | hahahahaha | 14:46 |
| TheJulia | BadCub: Clearly we need to get you a magical coffee mug | 14:46 |
| jroll | uh | 14:46 |
| jroll | wat | 14:46 |
| jroll | devananda: WAT | 14:47 |
| rfchapman | NobodyCam: OK i'll give it a try. | 14:47 |
| Shrews | O.o | 14:47 |
| *** achanda has joined #openstack-ironic | 14:47 | |
| devananda | jroll: apparently "db01.ironic" is an invalid name, but "unicorn" works fine | 14:47 |
| jroll | devananda: I quit. | 14:47 |
| BadCub | TheJulia: I was thinking of getting two coffee pots. I could just carry a pot around with me with a straw in it! | 14:47 |
| jroll | devananda: idk, passes the RFC hostname for me | 14:47 |
| jroll | er, hostname rfs | 14:47 |
| jroll | rfc too | 14:47 |
| jroll | ..... there's no test for anything with '.' | 14:48 |
| Shrews | m = '^[a-z0-9]([a-z0-9\-]{0,61}[a-z0-9])?$' | 14:48 |
| * jroll fixes this | 14:48 | |
| *** kkoski has joined #openstack-ironic | 14:48 | |
| devananda | jroll: http://git.openstack.org/cgit/openstack/ironic/tree/ironic/common/utils.py#n180 | 14:48 |
| devananda | yah .... | 14:48 |
| jroll | devananda: right. | 14:48 |
| jroll | BUT THE SPEC SAID NO DOTS | 14:48 |
| jroll | so it's ok. | 14:49 |
| devananda | exactly | 14:49 |
| * jroll looks at http://specs.openstack.org/openstack/ironic-specs/specs/kilo/logical-names.html#what-is-a-logical-name | 14:49 | |
| devananda | well ... actually a hostname can't have dots because those are the separator for FQDN | 14:49 |
| devananda | right? | 14:49 |
| rameshg87 | jroll: https://review.openstack.org/#/c/164313/7/nova/virt/ironic/driver.py | 14:49 |
| rameshg87 | jroll: L206 | 14:49 |
| rameshg87 | jroll: it should be not in, right ? | 14:49 |
| rameshg87 | node_obj.provision_state not in bad_provision_states | 14:50 |
| * jroll looks at http://tools.ietf.org/html/rfc952 | 14:50 | |
| jroll | rameshg87: doh, ty | 14:50 |
| rameshg87 | jroll: i guess it's doing exactly reverse of what is required | 14:50 |
| rameshg87 | :) | 14:50 |
| jroll | rameshg87: pushed a fix | 14:50 |
| jroll | right..... | 14:51 |
| jroll | now. | 14:51 |
| rameshg87 | okay :) | 14:51 |
| rameshg87 | just checking other things .. | 14:51 |
| rameshg87 | i was trying to make sense of that line in the previous patch assuming i am missing something | 14:51 |
| *** achanda has quit IRC | 14:52 | |
| Nisha | rloo, hi | 14:53 |
| Nisha | rloo, i raised the follow up patch for capabilities test cases https://review.openstack.org/#/c/165310/ | 14:53 |
| *** stendulker has joined #openstack-ironic | 14:54 | |
| Nisha | rloo, vdrok has some comments on the added test case. let me know if i have understood correct. | 14:54 |
| openstackgerrit | Jim Rollenhagen proposed openstack/ironic: Allow periods (".") in hostnames https://review.openstack.org/165443 | 14:54 |
| jroll | devananda: ^^ | 14:54 |
| devananda | hahahahah | 14:54 |
| *** kkoski has quit IRC | 14:55 | |
| BadCub | Nisha: is 165310 required to close the BP now? | 14:56 |
| jroll | devananda: had to restrain myself from words like "insanity" and "RFCs be damned" | 14:56 |
| * Shrews now knows that jroll likes name his pets with periods | 14:56 | |
| devananda | jroll: http://paste.openstack.org/show/193145/ | 14:56 |
| jroll | :D | 14:56 |
| jroll | devananda: hey, I can still write code, cool | 14:57 |
| Nisha | No | 14:57 |
| rfchapman | NobodyCam: I keep getting an error - http://paste.openstack.org/show/193146/ | 14:57 |
| Nisha | BadCub, no that is just minor test cases things | 14:57 |
| BadCub | Nisha: awesomeness :-) | 14:57 |
| Nisha | :) | 14:57 |
| Nisha | but first two patches are still remaining to be approved and merged | 14:57 |
| NobodyCam | rfchapman: looking | 14:58 |
| *** bauzas has joined #openstack-ironic | 14:58 | |
| BadCub | Nisha: One is +A already. Lucas is reviewing the other. | 14:58 |
| Nisha | ok | 14:58 |
| rfchapman | NobodyCam: thanks | 14:59 |
| Nisha | BadCub, need one more +2/+A on https://blueprints.launchpad.net/ironic/+spec/automate-uefi-bios-iso-creation | 15:00 |
| Nisha | #link https://review.openstack.org/155900 | 15:00 |
| Nisha | BadCub, ^^^^ | 15:00 |
| NobodyCam | rfchapman: humm.. looks like tripleo folks made some changes: did you see this line: WARNING: map-packages is deprecated. Please use the pkg-map element. | 15:00 |
| BadCub | Nisha: I am going down the list today. We are trying to hit as much as we can :-) | 15:00 |
| Nisha | BadCub, yeah i saw | 15:01 |
| BadCub | :-) | 15:01 |
| rfchapman | NobodyCam: Yes, I have been seeing that for some time now. | 15:04 |
| devananda | JoshNang: https://review.openstack.org/#/c/164954/ in good shape now, or are we still broken? | 15:05 |
| NobodyCam | rfchapman: give me several minutes to wrap up what I'm looking at | 15:05 |
| devananda | *is the agent_ssh job still ... | 15:05 |
| NobodyCam | rfchapman: also have you pinged anyone in the #TripleO channel? | 15:06 |
| rloo | Nisha: replied in the patch | 15:06 |
| *** romcheg has quit IRC | 15:06 | |
| *** romcheg1 has joined #openstack-ironic | 15:06 | |
| Nisha | rloo, ok i will just see | 15:07 |
| rfchapman | NobodyCam: Ahh good idea :-) | 15:07 |
| NobodyCam | rfchapman: :) | 15:07 |
| *** Marga_ has joined #openstack-ironic | 15:07 | |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: follow up patch for ilo capabilities https://review.openstack.org/165310 | 15:07 |
| Nisha | rloo, ^^^ | 15:08 |
| NobodyCam | rloo: i dont understand the comment at line 240 of: https://review.openstack.org/#/c/151596/45/ironic/drivers/modules/ilo/inspect.py | 15:08 |
| rloo | Nisha: I was purposely not going to look at that followup patch in case it conflicted with your other patch. so was going to wait. | 15:08 |
| NobodyCam | unless the patch Nisha just posted has somehting about it | 15:08 |
| rloo | NobodyCam: capabilities are done in the next patch, so it isn't true in that patch. | 15:08 |
| jroll | devananda: josh got that stuff working locally, I just fixed up the nova patches so we should be good now? I think/hope? | 15:09 |
| rloo | NobodyCam: but she copy/pasted from the other patch cuz I asked her about it. Not worth asking her to delete it here and add it in the next patch. | 15:09 |
| NobodyCam | ack!! Ty rloo :) | 15:09 |
| rloo | NobodyCam: I'll clarify in case others wonder. | 15:09 |
| jroll | devananda: 06:19:10 JoshNang | INFO ironic.conductor.manager [-] Node 0095c1b0-5dc8-48d5-9c1b-c56be951c43a cleaning complete | 15:09 |
| NobodyCam | jroll: :) nice | 15:09 |
| Nisha | rloo, it will not conflict as i understand. But i am fine if its done after first two patches | 15:10 |
| devananda | jroll: ack. i still see the fail on his patch, though | 15:10 |
| devananda | jroll: if the devstack change is done and correct, we should get that landed | 15:10 |
| Shrews | devananda: cleaning_network_uuid? will this work w/o neutron? | 15:10 |
| jroll | devananda: I just fixed the nova stuff | 15:11 |
| devananda | Shrews: yes. but it doesn't work with neutron unless you specify the network uuid | 15:11 |
| jroll | devananda: though that may not be relevant, unsure | 15:11 |
| devananda | jroll: oh. lemme check | 15:11 |
| devananda | so https://review.openstack.org/#/c/164954/ passes, but previous revs did anyway, because we're not running agent_ssh against devstack | 15:12 |
| devananda | it has 1 +2. when we're sure, i'll go poke in -qa to get it landed | 15:12 |
| NobodyCam | BadCub: /me just approved 151596 | 15:14 |
| BadCub | NobodyCam: okay | 15:14 |
| jroll | devananda: gimme a few and I'll check some stuff out | 15:15 |
| devananda | ugh. why am i seeing this spewing in the logs? INFO urllib3.connectionpool [-] Starting new HTTP connection | 15:15 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Suppress urllib3.connection INFO level logging https://review.openstack.org/165455 | 15:19 |
| *** pcaruana has joined #openstack-ironic | 15:19 | |
| devananda | jroll: you're right - network stuff in ironic worked, then it failed on the state in nova: http://logs.openstack.org/53/161453/12/check/check-tempest-dsvm-ironic-agent_ssh/63c1748/logs/screen-n-cpu.txt.gz?level=INFO#_2015-03-18_07_12_56_614 | 15:20 |
| NobodyCam | devananda: I thought we fixed that:/ | 15:20 |
| NobodyCam | brb | 15:23 |
| * BadCub goes to get more coffee | 15:24 | |
| devananda | JoshNang: also, we should be logging INFO messages for the cleaning steps | 15:24 |
| devananda | jroll: on https://review.openstack.org/#/c/164313/8 keep in mind that nova is going to see NOSTATE right now, because client is not passing any version header .... | 15:28 |
| *** lazy_prince is now known as killer_prince | 15:29 | |
| jroll | devananda: right, that's what I fixed | 15:29 |
| jroll | see line 203 | 15:29 |
| *** rameshg87 is now known as rameshg87-dinner | 15:30 | |
| *** penick has quit IRC | 15:33 | |
| *** kkoski has joined #openstack-ironic | 15:33 | |
| *** kkoski has quit IRC | 15:34 | |
| *** kkoski has joined #openstack-ironic | 15:34 | |
| devananda | jroll: ah ha | 15:38 |
| jroll | :) | 15:38 |
| devananda | just finished my morning sync with ttx | 15:38 |
| jroll | that patch originally failed waiting for resources | 15:38 |
| jroll | and? | 15:38 |
| JoshNang | good morning, sorry for the delay | 15:39 |
| devananda | tldr; he agrees that landing the nova-side fix of https://bugs.launchpad.net/ironic/+bug/1174153 during RC period is quite reasonable | 15:39 |
| openstack | Launchpad bug 1174153 in OpenStack Compute (nova) "data from previous tenants accessible with nova baremetal" [High,In progress] - Assigned to Jim Rollenhagen (jim-rollenhagen) | 15:39 |
| jroll | cool | 15:39 |
| NobodyCam | mornign JoshNang :) | 15:39 |
| JoshNang | nice | 15:39 |
| jroll | JoshNang: yeah what took you so long, it's been 9 hours since you left | 15:39 |
| jroll | geez | 15:39 |
| JoshNang | :P | 15:39 |
| devananda | jroll, JoshNang: so let's assume that nova isn't going to land https://review.openstack.org/#/c/164313 today | 15:39 |
| devananda | I think we need to land cleaning disabled for now | 15:39 |
| devananda | and enable it once they have accepted the fixes | 15:40 |
| devananda | because there is literally no other way to land it | 15:40 |
| jroll | yeah, seems fair | 15:40 |
| JoshNang | that's totally reasonable | 15:40 |
| jroll | the config already landed, we'll have to make another patch | 15:40 |
| devananda | we can't get it in turned on in the gate until nova accepts the fixes | 15:40 |
| *** absubram has quit IRC | 15:40 | |
| lucasagomes | :( | 15:40 |
| jroll | we could make a big depends-on change, but then that slows us down | 15:40 |
| jroll | I would love to see this pass the gate once before we turn that config off | 15:41 |
| devananda | JoshNang: you should, however, make another patch that turns it on, and then use that to test the feature on OTHER patches .... | 15:41 |
| jroll | or maybe have a second patch to turn iton... yeah | 15:41 |
| devananda | it's kinda awkward, but yea | 15:41 |
| JoshNang | devananda: that would work | 15:41 |
| lucasagomes | but if we land w/o nova, if someone enable it in Ironic it will fail as well right? | 15:41 |
| devananda | lucasagomes: ye | 15:41 |
| lucasagomes | we better make sure that the help message or something in ironic makes it clear | 15:41 |
| devananda | yes | 15:41 |
| devananda | mordred: sanity check me, if you're around | 15:42 |
| jroll | devananda: you want to draft that patch to disable it, with an awesome help string? | 15:42 |
| devananda | mordred: we have a 3-way dependency right now between feature in ironic, config option in devstack, and bug fix in nova | 15:42 |
| mordred | devananda: ok. | 15:43 |
| devananda | mordred: my plan: land config in devstack, then land feature disabled in ironic, propose patch to enable it tagged depends-on the nova change | 15:43 |
| JoshNang | fwiw, i used the devstack patch when i tested cleaning working last night, so it's definitely working | 15:43 |
| devananda | mordred: iterate as needed with nova (while we look at the ironic depends-on-nova change for results) until they accept it | 15:43 |
| devananda | JoshNang: yea, I saw that. good job | 15:43 |
| JoshNang | thanks! | 15:43 |
| devananda | JoshNang: that's actually what gives me the idea I just pitched to mordred | 15:44 |
| mordred | devananda: what does the devstack config do without the patches landed | 15:44 |
| mordred | ? | 15:44 |
| devananda | mordred: nothing. it adds a config option to ironic that we'll ignore | 15:44 |
| mordred | I assume it's safe to land that patch in general? | 15:44 |
| mordred | awesome | 15:44 |
| mordred | then yes, I think that your plan is correct | 15:44 |
| devananda | awesome, ty | 15:44 |
| jroll | +1, seems fine to me | 15:44 |
| devananda | jroll: /me drafts patch | 15:44 |
| BadCub | sounds like a whole lot of awesome to me | 15:45 |
| jroll | thanks dude | 15:45 |
| mordred | devananda: fwiw, you can also have multiple depends-on - so you could mark the ironic change depends on nova and depends on devstack | 15:45 |
| *** stendulker has quit IRC | 15:45 | |
| devananda | oh neat | 15:45 |
| mordred | but I don't think it's necessary | 15:45 |
| JoshNang | oh nice | 15:45 |
| jroll | mordred: devananda: though I think it would be nice to land this stuff for k3, even if nova takes longer. | 15:45 |
| devananda | jroll: I think we can land all the devstack and ironic things now | 15:46 |
| jroll | yeah | 15:46 |
| devananda | jroll: except for the "make it enabled b ydefault" | 15:46 |
| jroll | yep | 15:46 |
| jroll | agree | 15:46 |
| devananda | JoshNang: you have a bunch of comments on https://review.openstack.org/#/c/161453/12 | 15:46 |
| devananda | JoshNang: also the logging will need some cleanup there | 15:46 |
| JoshNang | i saw that. i'll get started as soon as i get coffee. yeah can do on the logging, i agree | 15:47 |
| devananda | cool | 15:47 |
| JoshNang | that patch really grew larger as i found bugs in the previous cleaning patch | 15:47 |
| JoshNang | but i kept them the same to make landing quicker. not sure if that's the right way to go about it | 15:47 |
| devananda | JoshNang: smaller patches are much easier to review in general .... | 15:48 |
| JoshNang | devananda: true. i can break them up while i fix comments. | 15:49 |
| devananda | k k | 15:49 |
| jlvillal | Good morning Ironic | 15:50 |
| NobodyCam | morning jlvillal :) | 15:50 |
| BadCub | devananda: while you guys are heads down on this, should we move forward with reviews on lower priority BPs? Or do you guys needs available eyes for these patches as they come up? | 15:50 |
| jroll | we have work to do, move forward! | 15:50 |
| BadCub | works for me :-) | 15:51 |
| * jlvillal feels like the yesterday was consumed with learning how to use select & fcntl to be able to capture the output from 'git fetch' and making a patch for repo. My past life is catching up with me ;) | 15:51 | |
| *** Nisha has quit IRC | 15:51 | |
| jlvillal | s/the yesterday/yesterday/ Not enough coffee yet... | 15:52 |
| devananda | BadCub: if ya'll can continue on the lower pri BP's that's great | 15:52 |
| *** kkoski has quit IRC | 15:53 | |
| * jroll reviews ilo cleaning stuff | 15:53 | |
| devananda | BadCub: getting reviews on the ilo code would be awesome | 15:53 |
| devananda | jroll: woot! ty | 15:53 |
| BadCub | ILO cleaning was my first item :-) | 15:53 |
| BadCub | anyone free to gander at https://review.openstack.org/#/c/157715/ and https://review.openstack.org/#/c/164646/ ? | 15:55 |
| dtantsur | JoshNang, just confirming: are you ok with https://review.openstack.org/#/c/157715/18/ironic/common/exception.py introducing NodeCleaningFailure too? | 15:55 |
| jroll | JoshNang: this is interesting, curious what you think https://review.openstack.org/#/c/164646/3/ironic/drivers/modules/ilo/deploy.py | 15:56 |
| JoshNang | dtantsur: i think i'm also adding one of my patches | 15:56 |
| dtantsur | JoshNang, that's why I ask | 15:56 |
| dtantsur | you definitely do | 15:56 |
| JoshNang | i mean, rebasing isn't hard :) | 15:57 |
| dtantsur | JoshNang, in this case it's risking having 2 copies of exception | 15:57 |
| JoshNang | or i'll break that out and hopefully land it today? | 15:57 |
| JoshNang | ahh | 15:57 |
| * BadCub has to go listen in on a call n a few minutes | 15:58 | |
| dtantsur | JoshNang, if you break your patch in generic and less generic ones, it would be sane to ask ramineni to rebase on top, but it should be somehow agreed | 15:58 |
| devananda | JoshNang: your comment on cleaning says "when preparing a node for deployments or after an instance is deleted" -- this seems wrong to me | 15:58 |
| *** kkoski has joined #openstack-ironic | 15:59 | |
| JoshNang | devananda: yeah that wording has changed like 3 times. i originally had as something like 'after instance is deleted or when moving from MANAGEABLE to AVAILABLE' | 15:59 |
| JoshNang | dtantsur: yeah, i have to make some generic patches this morning already, i can add that exception | 16:00 |
| *** ijw has joined #openstack-ironic | 16:00 | |
| devananda | oh -- btw folks -- approximately 24 hours from now is the cut off point, unless we delay the milestone. | 16:00 |
| dtantsur | JoshNang, please leave a comment on https://review.openstack.org/#/c/157715 then | 16:01 |
| * jroll approves everything | 16:01 | |
| JoshNang | dtantsur: will do | 16:01 |
| devananda | JoshNang: so "hopefully land it today" is an interesting statement :) | 16:01 |
| JoshNang | s/hopefully// | 16:01 |
| devananda | JoshNang: if something isn't really really close to ready to land now, we should defer it to liberty | 16:01 |
| devananda | jroll: :-D | 16:02 |
| *** ijw has quit IRC | 16:02 | |
| *** ijw has joined #openstack-ironic | 16:02 | |
| devananda | JoshNang: is CLEAN actually done between manageable -> available?? | 16:02 |
| ramineni | dtantsur: If 157715 gets merged before , JoshNang can always remove that file from his patch though ..:) | 16:02 |
| JoshNang | devananda: it is | 16:02 |
| dtantsur | ramineni, sure, I just want you to sync your efforts :) | 16:03 |
| JoshNang | ramineni: yup agreed | 16:03 |
| devananda | JoshNang: oooh 273 machine.add_transition(MANAGEABLE, CLEANING, 'provide') | 16:03 |
| *** stendulker has joined #openstack-ironic | 16:03 | |
| devananda | JoshNang: I missed that :( | 16:03 |
| *** kkoski has quit IRC | 16:03 | |
| JoshNang | yup | 16:03 |
| devananda | great | 16:03 |
| devananda | i mean, that's what the spec says :) | 16:03 |
| devananda | cheers | 16:04 |
| JoshNang | the reasoning was 'if cleaning prepares it for available after a node, it should prepare it coming out of manageable, especially the first time' | 16:04 |
| JoshNang | heh yup | 16:04 |
| devananda | totally. i just didn't notice that in the changelogs. all is good | 16:04 |
| JoshNang | gotcha | 16:04 |
| *** ijw has quit IRC | 16:04 | |
| *** ijw has joined #openstack-ironic | 16:05 | |
| *** jgrimm_ has joined #openstack-ironic | 16:07 | |
| *** maurosr- has joined #openstack-ironic | 16:08 | |
| *** mjturek2 has joined #openstack-ironic | 16:08 | |
| *** maurosr has quit IRC | 16:09 | |
| *** maurosr| has joined #openstack-ironic | 16:09 | |
| *** jgrimm- has joined #openstack-ironic | 16:09 | |
| *** maurosr| is now known as maurosr | 16:09 | |
| *** mjturek3 has joined #openstack-ironic | 16:10 | |
| *** jgrimm has quit IRC | 16:10 | |
| *** mjturek1 has quit IRC | 16:11 | |
| NobodyCam | BadCub: +2'd 154808 .. with a request for some corrections in a follow up patch | 16:12 |
| *** mjturek2 has quit IRC | 16:12 | |
| *** maurosr- has quit IRC | 16:12 | |
| BadCub | NobodyCam: awesome, TY :-) | 16:13 |
| *** jgrimm_ has quit IRC | 16:13 | |
| *** romcheg1 has quit IRC | 16:15 | |
| *** ramineni has quit IRC | 16:19 | |
| *** ramineni1 has joined #openstack-ironic | 16:19 | |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Disable cleaning by default https://review.openstack.org/165241 | 16:19 |
| devananda | JoshNang: ^ | 16:19 |
| *** killer_prince has quit IRC | 16:19 | |
| *** Nisha has joined #openstack-ironic | 16:20 | |
| *** killer_prince has joined #openstack-ironic | 16:20 | |
| *** killer_prince is now known as lazy_prince | 16:20 | |
| JayF | JoshNang: if you have run shred on a node, can you make the test output on my IPA fix better? Or gist the output from shred and I'll do it? | 16:20 |
| JayF | JoshNang: literally right now I have utils.execute returning "shred output" to the mock'd stdout | 16:21 |
| JayF | lol | 16:21 |
| JoshNang | devananda: reviewed! | 16:23 |
| *** rameshg87-dinner is now known as rameshg87 | 16:23 | |
| rameshg87 | ramineni1: are you around ? | 16:23 |
| ramineni1 | rameshg87: yes | 16:23 |
| JoshNang | JayF: so that's what was returned as an exception because we're doing 'exception was: str(e)' somewhere. i tested changing the passes to str(passes) and it worked in my vm | 16:25 |
| JoshNang | sorry, i would have fixed it last night but it was already...quite late | 16:25 |
| rameshg87 | JoshNang, ramineni1, can we define a single option for defining priority of erase devices | 16:25 |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: [WIP] Add DevStack plugin for ironic-discoverd https://review.openstack.org/164782 | 16:25 |
| dtantsur | hacking devstack is "fun" >_< | 16:25 |
| rameshg87 | JoshNang: ramineni1: we have CONF.ilo.clean_priority_erase_devices defined in https://review.openstack.org/#/c/164646/3/ironic/drivers/modules/ilo/deploy.py | 16:25 |
| rfchapman | NobodyCam: No one on the #tripleo channel answered, so I did the radical thing and removed the element that was giving me a problem. 10-cloud-init. Actually I didn't remove the entire element, just one of the install scripts that was specific to ubuntu. | 16:26 |
| rameshg87 | JoshNang: ramineni1: and CONF.agent.agent_erase_devices_priority defined in https://review.openstack.org/#/c/161453/12/ironic/drivers/modules/agent.py | 16:26 |
| jroll | rameshg87: hey, does ilo_password and ipmi_password need to be the same for ilo drivers? (iirc we use IPMI for sensor data there) | 16:26 |
| jroll | rameshg87: ^ | 16:26 |
| jroll | ramineni1: ** ^ | 16:26 |
| rameshg87 | JoshNang: ramineni1: i think it should just be CONF.deploy.clean_priority_erase_devices | 16:26 |
| rameshg87 | jroll: yes | 16:26 |
| jroll | rameshg87: boo, this patch was so close. thanks. | 16:26 |
| *** kkoski has joined #openstack-ironic | 16:26 | |
| NobodyCam | rfchapman: ya I saw: dd: failed to open '/etc/cloud/cloud.cfg.d/10_etc_hosts.cfg': No such file or directory | 16:27 |
| dtantsur | going a bit earlier, see you tomorrow | 16:27 |
| *** kkoski has quit IRC | 16:27 | |
| ramineni1 | rameshg87: thought of using the same, but added to be consistent with other clean options | 16:27 |
| *** dtantsur is now known as dtantsur|afk | 16:27 | |
| NobodyCam | night dtantsur|afk | 16:27 |
| *** kkoski has joined #openstack-ironic | 16:27 | |
| rameshg87 | ramineni1: but it is one and the same | 16:27 |
| rameshg87 | ramineni1: it's cleaning with agent - just that it is happening through a different driver which boots agent in a different way | 16:28 |
| rameshg87 | ramineni1: operation is still the same :) | 16:28 |
| rameshg87 | ramineni1: and it happens in the same agent, the same call :) | 16:28 |
| jroll | ramineni1: one thing here when you have time https://review.openstack.org/#/c/157715/ | 16:28 |
| *** lazy_prince has quit IRC | 16:29 | |
| JoshNang | yeah, one config option is probably easier | 16:29 |
| JoshNang | well, not easier for us, easier for the operators | 16:29 |
| rameshg87 | jroll: but we don't need to define ipmi_password https://github.com/openstack/ironic/blob/master/ironic/drivers/modules/ilo/common.py#L203-L205 | 16:29 |
| ramineni1 | rameshg87: i'm ok with it .. | 16:29 |
| jroll | rameshg87: ramineni1: if you have some ilo nodes and some non-ilo nodes, it might make sense for the erase devices priority to be different between those no? | 16:29 |
| rfchapman | NobodyCam: It should eather create the directory structure if it doesn't exist or check to make sure it's ubuntu that is being built. | 16:29 |
| rameshg87 | jroll: the code automatically updates ipmi_password but doesn't save | 16:29 |
| jroll | rameshg87: OH | 16:29 |
| NobodyCam | rfchapman: also may be worth looking at https://github.com/openstack/diskimage-builder/tree/master/elements/cloud-init-nocloud element | 16:29 |
| jroll | rameshg87: thank you | 16:29 |
| jroll | ramineni1: BadCub: https://review.openstack.org/#/c/157715/ +A | 16:30 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Disable cleaning by default https://review.openstack.org/165241 | 16:30 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Enable cleaning by default https://review.openstack.org/165496 | 16:30 |
| ramineni1 | jroll: why is that? didnt get your point | 16:30 |
| rameshg87 | jroll: pets vs cattles :) | 16:30 |
| *** kkoski has quit IRC | 16:30 | |
| *** killer_prince has joined #openstack-ironic | 16:30 | |
| BadCub | jroll: awesome! TY!!! | 16:30 |
| devananda | JoshNang: fixed flake8 error, and now with the depends-on patch | 16:30 |
| JoshNang | devananda: sweet | 16:30 |
| *** killer_prince is now known as lazy_prince | 16:30 | |
| ramineni1 | jroll: great ..thanks :) | 16:30 |
| jroll | ramineni1: I fixed it :) | 16:30 |
| * devananda re coffees | 16:31 | |
| jroll | rameshg87: no, maybe my Acme brand servers need bios updated before erasing devices or something. | 16:31 |
| JoshNang | jroll: rameshg87 yeah, i guess with multiple drivers it makes sense to have them at different priorities. a single node will never have both drivers (both deploy) | 16:31 |
| *** dlpartain has joined #openstack-ironic | 16:31 | |
| jroll | rameshg87: IMO it's reasonable to have separate configs for separate drivers (though it sucks) | 16:31 |
| stendulker | BadCub: Hi | 16:31 |
| rameshg87 | JoshNang: jroll: okay :) | 16:31 |
| JoshNang | yeah :/ | 16:32 |
| rameshg87 | ramineni1: JoshNang: then ignore that :) | 16:32 |
| stendulker | BadCub: In IronicReviewDay etherpad, pxe secure boot patch https://review.openstack.org/#/c/154808/ maked as 'Approved' | 16:32 |
| rfchapman | NobodyCam: will that cause problems with OpenStack? | 16:32 |
| *** dlpartain has left #openstack-ironic | 16:32 | |
| rfchapman | It only mentioned EC2 | 16:32 |
| stendulker | BadCub: It needs another +2/+A. May need a correction in etherpad | 16:33 |
| ramineni1 | rameshg87: ok | 16:33 |
| NobodyCam | that i believe kills cloud-init. | 16:33 |
| NobodyCam | rfchapman: do you have a meta data service running? | 16:33 |
| * ramineni1 working on other comments on the patch | 16:33 | |
| BadCub | corrected | 16:33 |
| stendulker | BadCub: Thank you :) | 16:33 |
| BadCub | stendulker: :-) | 16:34 |
| Shrews | JoshNang, and others: I think we may be doing this wrong, with respect to the message used with cleaning_error_handler(), but I need a sanity check: https://github.com/openstack/ironic/blob/master/ironic/conductor/manager.py#L921-L926 | 16:34 |
| rfchapman | NobodyCam: Not that I'm aware of... just basic OpenStack-ironic (PACKSTACK) | 16:34 |
| Shrews | JoshNang: I *think* that message should be wrapped with _LE() since it isn't being raised within any exception | 16:35 |
| jroll | Shrews: that message is put in node.last_error, where it isn't translated | 16:35 |
| Shrews | jroll: ah, yes. i glossed over that, apparently. thanks. | 16:36 |
| jroll | :) | 16:36 |
| Shrews | sanity restored (temporarily) | 16:36 |
| jroll | what sanity? | 16:36 |
| devananda | rfchapman: fwiw, i dont know if any of the developers here have used packstack w/ ironic ... | 16:36 |
| * jroll ducks | 16:36 | |
| * BadCub now finished with call | 16:36 | |
| jroll | devananda: rfchapman: some developers here don't even know what packstack is | 16:36 |
| devananda | heh | 16:37 |
| JoshNang | Shrews: heh yeah. the whole LE() thing is really confusing when using the message anywhere other than an exception. i have to check the docs to sanity check pretty often | 16:37 |
| *** lazy_prince has quit IRC | 16:37 | |
| rfchapman | devananda: I'm kinda finding this out late in the game. Unfortunatly it was the only distro that I could get to work. | 16:37 |
| devananda | rfchapman: packstack is a distro? | 16:38 |
| *** killer_prince has joined #openstack-ironic | 16:38 | |
| *** killer_prince is now known as lazy_prince | 16:38 | |
| rfchapman | devananda: maybe distro was the wrong word to use, software stack I guess is better :-) | 16:39 |
| devananda | JoshNang: should https://review.openstack.org/#/c/161453/ also depends-on the nova changes? | 16:40 |
| devananda | JoshNang: wait. never mind | 16:40 |
| JoshNang | no worries. i think that'd create a cycle | 16:41 |
| JayF | JoshNang: I think you misunderstood what I asked earlier, but you're busy so keep going :) | 16:41 |
| JoshNang | JayF: oh sorry :/ | 16:41 |
| JayF | JoshNang: I'm saying, you ran shred on a thing in devstack. Do you have the logged output from the shred command? | 16:41 |
| JoshNang | ohh i see | 16:42 |
| JoshNang | lemme check my logs! | 16:42 |
| jroll | devananda: don't we want to land this bit now and -2 the "enable cleaning" patch? https://review.openstack.org/#/c/165241/ | 16:42 |
| devananda | jroll: yup. i was just fixing those votes | 16:42 |
| *** Marga_ has quit IRC | 16:42 | |
| jlvillal | I think the 'E' in _LE stands for 'error'. Not sure if it is used for both LOG.error and LOG.exception though. | 16:43 |
| *** Marga_ has joined #openstack-ironic | 16:43 | |
| devananda | jroll: also we shouldn't need a -2 on https://review.openstack.org/#/c/165496/ because it should just fail tests until then | 16:43 |
| jroll | devananda: cool, thanks | 16:43 |
| jroll | other cores can we get a +A on https://review.openstack.org/#/c/165241/ | 16:43 |
| rfchapman | devananda: I tried a couple and even had problems getting devstack running behind the proxy. I installed it on my server at home and brought the image of the VM in. At home I had no problem installing packstack. I also tried DevStack at home with out luck. | 16:43 |
| jroll | devananda: good point :) | 16:43 |
| devananda | jroll: but yah. I think we *have to* land 165241 now | 16:43 |
| Shrews | jroll: just looking at that | 16:43 |
| jroll | Shrews: doeeeeeeeeeet | 16:43 |
| rameshg87 | jroll: so decided ? | 16:43 |
| JoshNang | jlvillal: yup exception and error. http://docs.openstack.org/developer/oslo.i18n/guidelines.html | 16:43 |
| jlvillal | JoshNang: Thanks! | 16:44 |
| * rameshg87 reads back | 16:44 | |
| jroll | rameshg87: ? | 16:44 |
| jlvillal | JoshNang: and bookmarked :) | 16:44 |
| rameshg87 | jroll: decided that nova patch is not going in ? | 16:44 |
| jroll | rameshg87: we're diabling cleaning by default until the nova changes land | 16:44 |
| JayF | JoshNang: responded here https://review.openstack.org/#/c/165247/2/ironic_python_agent/hardware.py,cm | 16:44 |
| JoshNang | jlvillal: heh yeah i've checked that a bunch of times | 16:44 |
| devananda | rfchapman: huh. devstack works for all of us ... another time I'd like to know what you had problems with | 16:44 |
| jroll | rameshg87: nova says they will accept it | 16:44 |
| JayF | JoshNang: I'll fix the thing (wrap it in six.text_type() ) -- but curious about your opinion on the try: except: thing you suggested | 16:45 |
| rameshg87 | jroll: oh okay .. | 16:45 |
| devananda | rameshg87: jroll: well. they're not blocking a bug fix. but they hvaen't said "yes" yet either | 16:45 |
| jroll | devananda: sure, words are hard | 16:45 |
| Shrews | devananda: jroll: +A on 165241 | 16:45 |
| devananda | indeed | 16:45 |
| *** lazy_prince has quit IRC | 16:46 | |
| rfchapman | devananda: sure I'll see if I can dig up some notes on the errors I was having. | 16:46 |
| jroll | thanks Shrews | 16:46 |
| devananda | jroll: so the help string i wrote on 165241 should continue to be viable -- even if the fix does not land in nova in Kilo | 16:46 |
| NobodyCam | brb | 16:46 |
| devananda | and we are forced to release kilo with clean disabled | 16:46 |
| devananda | i'll be sad, but i think that message is informative enough | 16:46 |
| JoshNang | JayF: replied | 16:47 |
| *** penick has joined #openstack-ironic | 16:48 | |
| JayF | JoshNang: also as part of writing that patch, I found we do a lot of utils.execute() without check_exit_code=[0] | 16:48 |
| jroll | devananda: indeed, I liked it | 16:48 |
| devananda | JayF: eek | 16:48 |
| JoshNang | JayF: :( :( | 16:48 |
| JayF | JoshNang: I might whip up a patch to do that today throughout the GenericHardwareManager | 16:48 |
| JayF | low effort high value | 16:48 |
| jroll | JayF: I thought [0] was default | 16:48 |
| jroll | JayF: https://github.com/openstack/oslo.concurrency/blob/master/oslo_concurrency/processutils.py#L157 | 16:49 |
| JayF | aww hell | 16:49 |
| jroll | lol | 16:49 |
| JayF | I'll remove that from my change then | 16:49 |
| *** killer_prince has joined #openstack-ironic | 16:49 | |
| jroll | quite the sane default ;D | 16:49 |
| *** killer_prince is now known as lazy_prince | 16:49 | |
| JoshNang | JayF: (checked logs) i don't think the result of that shred is getting returned to ironic | 16:50 |
| JayF | JoshNang: ah, you can't get to the IPA logs? | 16:50 |
| JayF | I'll just have to shred a machine in the lab then :/ | 16:50 |
| JoshNang | JayF: nope. default ironic powers off the nodes after it does things ;) | 16:50 |
| JayF | Why can't all the Ironics in the world be configured like mine!? /s | 16:50 |
| JoshNang | heh | 16:51 |
| *** penick has quit IRC | 16:51 | |
| *** kkoski has joined #openstack-ironic | 16:52 | |
| *** kkoski has quit IRC | 16:53 | |
| *** kkoski has joined #openstack-ironic | 16:53 | |
| * Shrews decides to try this new fad called "lunch" | 16:55 | |
| *** absubram has joined #openstack-ironic | 16:55 | |
| NobodyCam | Shrews: you'll like it! | 16:55 |
| *** kkoski has quit IRC | 16:55 | |
| devananda | I'm bumping https://bugs.launchpad.net/ironic/+bug/1431204 to liberty as this is going to be both more work and non-trivial API changes | 16:56 |
| openstack | Launchpad bug 1431204 in Ironic "New field 'name' not supported in port REST API" [Medium,In progress] - Assigned to Michael Davies (mrda) | 16:56 |
| BadCub | brb | 16:58 |
| *** openstackgerrit has quit IRC | 16:59 | |
| *** ukalifon1 has quit IRC | 16:59 | |
| *** openstackgerrit has joined #openstack-ironic | 16:59 | |
| *** stendulker has quit IRC | 16:59 | |
| devananda | jroll: so https://review.openstack.org/#/c/165430/ passed all our tempest and integration tests .... :-/ | 17:00 |
| devananda | just like we both expected | 17:00 |
| devananda | jroll: what board do we need to pin this to to get the client fixed? | 17:00 |
| *** ramineni has joined #openstack-ironic | 17:01 | |
| *** romcheg has joined #openstack-ironic | 17:01 | |
| *** vdrok is now known as vdrok_afk | 17:01 | |
| jroll | devananda: fun. I have the same question | 17:01 |
| adam_g | devananda, is that a client bug or ? | 17:01 |
| *** alexpilotti has quit IRC | 17:02 | |
| *** alexpilotti_ has joined #openstack-ironic | 17:02 | |
| *** penick has joined #openstack-ironic | 17:02 | |
| *** MattMan has quit IRC | 17:02 | |
| *** ramineni1 has quit IRC | 17:02 | |
| devananda | jroll: https://review.openstack.org/#/c/155624/ | 17:02 |
| *** mgoddard has quit IRC | 17:03 | |
| devananda | jroll: I argued heavily about this on that patch, but meh, couldn't convince anyone else | 17:03 |
| devananda | adam_g: let's say ... it's debatable | 17:04 |
| * devananda fixes the client | 17:04 | |
| *** achanda has joined #openstack-ironic | 17:04 | |
| jroll | devananda: yeah, and I stayed out of it for whatever reason | 17:04 |
| devananda | JoshNang: anything else on cleaning i can help with right now? | 17:05 |
| devananda | JoshNang: I believe you're working on refactoring / incorporating feedback on the main agent patch, and so i dont want to touch that until you're done -- but point me at something else and i'll dive in | 17:05 |
| JoshNang | devananda: yup, the patch fixing bad task_manager handling in the conductor is almost done | 17:06 |
| devananda | cool | 17:06 |
| devananda | i'll go fix this client thing then, until you poke me agai | 17:06 |
| JoshNang | perfect | 17:07 |
| jroll | I'm also fairly free if folks need help with thingas | 17:07 |
| jroll | doing reviews otherwise | 17:07 |
| BadCub | thank you jroll :-) | 17:08 |
| *** jistr has quit IRC | 17:08 | |
| BadCub | thank you everyone! | 17:08 |
| NobodyCam | huh: ImportError: No module named oslo_policy | 17:12 |
| devananda | NobodyCam: that hit me too. refresh your venv | 17:12 |
| rameshg87 | NobodyCam: newly added, | 17:12 |
| NobodyCam | ahh | 17:12 |
| devananda | NobodyCam: or just pip install -U -r requiremets.txt | 17:12 |
| NobodyCam | adds -r to his tox command | 17:12 |
| NobodyCam | prob a good thing to rebuiild | 17:12 |
| jroll | NobodyCam: btw, if you aren't using it, this helps a ton: https://www.berrange.com/posts/2014/11/14/faster-rebuilds-for-python-virtualenv-trees/ | 17:13 |
| NobodyCam | oh neat.. ty jroll will look at that | 17:13 |
| jroll | np | 17:14 |
| *** bradjones has quit IRC | 17:14 | |
| *** adam_g has left #openstack-ironic | 17:16 | |
| *** adam_g has joined #openstack-ironic | 17:16 | |
| *** hj-hp has joined #openstack-ironic | 17:16 | |
| Shrews | devananda: i'm just going to push up a patch fixing my nit on https://review.openstack.org/158520 | 17:18 |
| openstackgerrit | David Shrewsbury proposed openstack/python-ironicclient: Add support for logical names https://review.openstack.org/158520 | 17:20 |
| devananda | Shrews: +1'd | 17:21 |
| Shrews | I'm still going to +2 it | 17:21 |
| *** penick has quit IRC | 17:22 | |
| *** ramineni1 has joined #openstack-ironic | 17:23 | |
| *** ramineni has quit IRC | 17:24 | |
| *** rsalevsky has quit IRC | 17:24 | |
| *** harlowja_away is now known as harlowja_ | 17:27 | |
| NobodyCam | brb | 17:28 |
| *** Marga_ has quit IRC | 17:28 | |
| *** ramineni has joined #openstack-ironic | 17:28 | |
| openstackgerrit | Josh Gachnang proposed openstack/ironic: Use task.spawn_after to maintain lock during cleaning https://review.openstack.org/165528 | 17:30 |
| *** ramineni1 has quit IRC | 17:30 | |
| JoshNang | ^ there's one of the bugs i hit testing agent cleaning last night | 17:30 |
| openstackgerrit | Merged openstack/ironic: Add pxe_irmc to the sending IPMI sensor data driver list https://review.openstack.org/165313 | 17:32 |
| JoshNang | jroll: those two nova patches are passing now \o/ | 17:33 |
| JoshNang | thanks :) | 17:33 |
| jroll | yay, I can still write code! | 17:33 |
| jroll | np :) | 17:33 |
| *** kkoski has joined #openstack-ironic | 17:36 | |
| rfchapman | NobodyCam: Do I need the (vm) option for the ramdisk-image-create? | 17:36 |
| NobodyCam | rfchapman: no you should not | 17:37 |
| rfchapman | It looks like the disk-image-create complete error free... Now building the deploy ramdisk and kernel. | 17:39 |
| openstackgerrit | Jim Rollenhagen proposed openstack/python-ironicclient: Add support for logical names https://review.openstack.org/158520 | 17:40 |
| JoshNang | i'm gonna step away for a few minutes then tackle comments on agent cleaning | 17:40 |
| jroll | devananda: Shrews: fixed a small nit (port list doesn't work with names) | 17:40 |
| * jroll +2's and walks away for a moment | 17:40 | |
| *** coolsvap is now known as coolsvap|afk | 17:41 | |
| JoshNang | though, i would love more eyes on the release_resources bit of https://review.openstack.org/#/c/161453/12/ironic/drivers/modules/agent_base_vendor.py | 17:41 |
| JoshNang | i'd love for there to be a better way (maybe i missed it) for a driver to notify the conductor it is done with an async op other than rpc'ing | 17:42 |
| *** romcheg has quit IRC | 17:42 | |
| * devananda looks | 17:43 | |
| devananda | JoshNang: oh god no | 17:43 |
| devananda | JoshNang: driver runs inside the conductor. there's no sane reason it needs to send an RPC message to itself ... | 17:45 |
| *** ramineni1 has joined #openstack-ironic | 17:47 | |
| *** ramineni has quit IRC | 17:47 | |
| devananda | JoshNang: so IIUC the problem is this: | 17:47 |
| devananda | - for some cleaning operations, the driver might wait for it to finish, then return to the conductor | 17:47 |
| JoshNang | devananda: correct | 17:48 |
| devananda | - for other operations, the driver does not wait, and so the conductor needs to be notified in some other way | 17:48 |
| JoshNang | right | 17:48 |
| openstackgerrit | Merged openstack/ironic: Common changes for secure boot support https://review.openstack.org/153974 | 17:48 |
| openstackgerrit | Merged openstack/ironic: Common changes for secure boot support https://review.openstack.org/153974 | 17:48 |
| devananda | maybe this is too simplistic, but the driver call should not be async | 17:48 |
| *** ParsectiX has joined #openstack-ironic | 17:49 | |
| *** wanyen has joined #openstack-ironic | 17:49 | |
| devananda | the RPC call (from api -> conductor) to start or to continue cleaning need to be sync and return 202 accepted when they start a worker thread | 17:49 |
| devananda | but that worker thread should not terminate until the driver is done cleaning | 17:49 |
| JoshNang | alright | 17:49 |
| devananda | *done with as much as it can do | 17:49 |
| *** derekh has quit IRC | 17:49 | |
| JoshNang | so...for the agent, it would tell teh agent to run a command, check the result, if still running, do whatever greenthread's sleep is? | 17:50 |
| JoshNang | wake up, recheck, etc? | 17:50 |
| JoshNang | (i'm totally fine with that, makes a lot of things simple) | 17:50 |
| devananda | JoshNang: there are discrete clean steps | 17:51 |
| devananda | JoshNang: it looks like resume between steps is already well hanled? | 17:51 |
| *** ChuckC has quit IRC | 17:51 | |
| *** ChuckC_ has joined #openstack-ironic | 17:51 | |
| JoshNang | correct | 17:51 |
| *** ChuckC_ is now known as ChuckC | 17:52 | |
| devananda | JoshNang: so within a single step, the conductor's worker thread should stay "alive" until that step is done | 17:52 |
| devananda | JoshNang: if the thread terminates early, the step may need to be resumed | 17:52 |
| devananda | or may hav eerrored | 17:52 |
| devananda | or what ever | 17:52 |
| devananda | s/resumed/retried/ | 17:52 |
| JoshNang | right | 17:52 |
| JayF | For long-running cleaning steps in large Ironic installs ... isn't that a way to run out of workers in a hurry? | 17:52 |
| devananda | yup | 17:53 |
| devananda | raise the # of workers | 17:53 |
| Shrews | jroll: well spotted | 17:53 |
| JayF | I know there are knobs for such things, just wanted to make sure I understood the downside | 17:53 |
| devananda | JayF: i'm not sure if turning that knob has other negative effects or not ... | 17:53 |
| JayF | I know in our installation we had to turn worker threads down at one point iirc | 17:54 |
| *** pas-ha has quit IRC | 17:55 | |
| JoshNang | devananda: so i think i'm still a bit confused. when the conductor is about to start executing a step, does it just call the step in the same thread? or spawn a new worker and end there? | 17:55 |
| *** ndipanov has quit IRC | 17:55 | |
| devananda | JoshNang: i think i'm confused about sometehing too | 17:56 |
| JoshNang | heh | 17:56 |
| Shrews | hrm, i recall some discussion around the number of worker threads a loooong time ago... something not good, iirc | 17:56 |
| devananda | JoshNang: doesn't the agent driver, once it has requested the agent to perform a given clean task, release the thread and wait for a callback to the API ? | 17:56 |
| devananda | JoshNang: that callback should get routed to the conductor and result in continuing the cleaning | 17:56 |
| lucasagomes | folks I will call it a day | 17:56 |
| lucasagomes | have a good night everyone! | 17:56 |
| devananda | JoshNang: so I dont see the need for agent_base_vendor to call rpc.continue_node_clean | 17:57 |
| NobodyCam | night lucasagomes | 17:57 |
| JoshNang | devananda: no. it pings back with a heartbeat | 17:57 |
| devananda | lucasagomes: cheers, g'night! | 17:57 |
| devananda | JoshNang: o.0 ? | 17:57 |
| *** lucasagomes is now known as lucas-dinner | 17:57 | |
| JoshNang | the driver checks the state of the last command. | 17:57 |
| BadCub | g'night lucas-dinner | 17:57 |
| JoshNang | if its still running, end, if its finished... ?? | 17:57 |
| JoshNang | that's how the rest of the agent commands work | 17:57 |
| JoshNang | continue node clean doesn't have an api endpoint (i thought it would be internal rpc only) | 17:58 |
| devananda | oh right | 17:58 |
| *** jcoufal has quit IRC | 18:02 | |
| devananda | JoshNang: so i thought we had agreed that cleaning would have an api-visiblw way to 'continue' it | 18:03 |
| JoshNang | devananda: did we? | 18:03 |
| devananda | that's needed for any outside-the-conductor tool to signal "hey, i'm done keep going" | 18:04 |
| JoshNang | devananda: hmm yeah that sounds familiar | 18:04 |
| jroll | I think today that is a vendor passthru, I don't remember agreeing on something like that | 18:04 |
| devananda | eg, if i'm using an in-band (but out of tree) method foe building a raid, i need to d othat when it's done | 18:04 |
| devananda | sure - but iirc that's all in the state mchine spec | 18:05 |
| openstackgerrit | Anusha Ramineni proposed openstack/ironic: Support agent_ilo driver to perform cleaning https://review.openstack.org/164646 | 18:05 |
| jroll | we certainly need something like that for zapping. do we need it for cleaning? | 18:05 |
| JoshNang | yeah i remember this more for zapping | 18:05 |
| devananda | jroll: apparently so | 18:05 |
| openstackgerrit | Chris Krelle proposed openstack/ironic: Check temp dir is writable for ipmitool driver https://review.openstack.org/160383 | 18:05 |
| jroll | well | 18:05 |
| devananda | because apparently the agent has to phone home multiple times between steps | 18:05 |
| jroll | it's going to be all in the driver | 18:05 |
| jroll | right now it's limited to one driver | 18:05 |
| jroll | so vendor passthru for now, right? | 18:06 |
| jroll | (or I should say vendor passthru is ok for now) | 18:06 |
| openstackgerrit | John Trowbridge proposed stackforge/ironic-discoverd: eDeploy: Store benchmark data from edeploy in Ironic DB https://review.openstack.org/165538 | 18:06 |
| NobodyCam | jlvillal: when you have a free minute.. can you see if my changes to 160383 are what you had in mind? | 18:06 |
| jroll | IOW do we want to figure out a new API endpoint one day before a release :P | 18:06 |
| jlvillal | NobodyCam: Will do! | 18:06 |
| devananda | jroll: no :) | 18:07 |
| devananda | jroll: vp ++ | 18:07 |
| JoshNang | ++ heh | 18:07 |
| jroll | :) | 18:07 |
| *** ParsectiX has quit IRC | 18:07 | |
| *** kkoski has quit IRC | 18:08 | |
| *** ParsectiX has joined #openstack-ironic | 18:08 | |
| JoshNang | so, how does the agent say "i'm done"? i agree internal rpc isn't great, but it's effectively the same as the agent hitting an api to say continue (except the agent driver gets a chance to check things beforehand) | 18:08 |
| wanyen | So if a driver uses non-IPA agent-based cleaning, the agent can call back using Ironic API through vendor pass-through? | 18:09 |
| NobodyCam | lol three +2's on 165443 (Allow periods (".") in hostnames) guess I should approve it :-p | 18:09 |
| jroll | wanyen: the driver would need to have that passthru call | 18:09 |
| jroll | JoshNang: ... through the heartbeat? I don't get it. | 18:09 |
| jroll | JoshNang: can the agent access the conductor manager? | 18:10 |
| JoshNang | hmm is it stored on task somewhere maybe? | 18:10 |
| devananda | jroll: in the current patch, agent driver is rleasing a lock then sending an RPC message | 18:10 |
| devananda | to itself | 18:10 |
| jroll | right, the release the lock part seems bad | 18:11 |
| jroll | I'm wondering if the agent could call task.manager.continue_clean or something | 18:11 |
| *** Marga_ has joined #openstack-ironic | 18:12 | |
| JoshNang | i don't know that task has an reference to manager | 18:12 |
| *** kkoski has joined #openstack-ironic | 18:12 | |
| JoshNang | does context? | 18:12 |
| devananda | JoshNang: if not, it's trivial to add that | 18:13 |
| devananda | JoshNang: and low risk. conversely, instantiating the RPC client in the same proces that is listening on the other end ... seems a little more extreme :p | 18:14 |
| JoshNang | :P | 18:14 |
| *** kkoski has quit IRC | 18:14 | |
| devananda | JoshNang: also, you're instantiating N clients, because you create a new one every time continue_cleaning is called | 18:14 |
| devananda | but they all hvae the same hostname | 18:14 |
| *** ijw has quit IRC | 18:15 | |
| devananda | actually nm, that's probably fine for the clients | 18:15 |
| *** igordcard has quit IRC | 18:16 | |
| *** ijw has joined #openstack-ironic | 18:16 | |
| *** tiagogomes_ has quit IRC | 18:17 | |
| devananda | actually forget all tht -- not trivial to add | 18:17 |
| devananda | JoshNang: something else bothers me here | 18:17 |
| wanyen | JoshNang, the bare-metal node does not need to be in power-on state when conductor calls driver to execute a cleaning step. right? | 18:17 |
| devananda | JoshNang: the API we have held drivers to is: manager calls driver, driver does $thing and returns status | 18:17 |
| devananda | JoshNang: you're changing this radically by allowing a driver to call the manager | 18:17 |
| JoshNang | wanyen: if its in band, yes. it'll be powered on by prepare_cleaning if necessary | 18:17 |
| devananda | JoshNang: but you're not actually changing the APi right now, because you send an RPC message instead | 18:18 |
| devananda | gah | 18:18 |
| wanyen | JoshNang, it depends onthe driver. If driver can do cleaning steps without pre-power on the bm node. | 18:18 |
| JoshNang | wanyen: correct | 18:18 |
| wanyen | JoshNang: +1 | 18:19 |
| jlvillal | NobodyCam: reviewed. Only a few minor nits in my opinion. | 18:19 |
| NobodyCam | jlvillal: awesome will look in a minute or so-ish | 18:19 |
| jroll | devananda: qq, is there a released version of the client that supports --version or whatever? | 18:19 |
| devananda | jroll: no. i haven't tagged a release since lintan's patch landed | 18:19 |
| jroll | devananda: ok, I was thinking on looking at adding tempest tests for other versions | 18:20 |
| devananda | ++ | 18:20 |
| jroll | but we need a release, I assume | 18:20 |
| devananda | jroll: also tempest API testing doesn't use our client | 18:20 |
| jroll | oh. | 18:20 |
| jroll | right. | 18:20 |
| jroll | ok, ignore me then | 18:20 |
| devananda | :) | 18:21 |
| devananda | jroll: we need a release before Nova will see the "AVAILABLE" state | 18:21 |
| devananda | in the integration tests | 18:21 |
| *** romcheg has joined #openstack-ironic | 18:21 | |
| jroll | devananda: yeah, and that release needs to use a later version | 18:21 |
| devananda | *a release where the default version is > 1.2 | 18:21 |
| devananda | right | 18:21 |
| * jroll wonders if we run tempest on client jobs | 18:22 | |
| devananda | also, i think the api service should log the version header | 18:22 |
| jroll | devananda: by the way I added a big "this is all broken" about this to the whiteboard gate status | 18:22 |
| devananda | jroll: we should run the integratoin test using it. <<< adam_g | 18:22 |
| devananda | jroll: great | 18:22 |
| devananda | jroll: i want the api log records, eg | 18:23 |
| devananda | 127.0.0.1 - - [18/Mar/2015 11:22:47] "GET /v1/nodes HTTP/1.1" 200 759 | 18:23 |
| devananda | to include the version header. becaues debugging this right now means i need to tcpdump | 18:23 |
| devananda | and i get grumpy when i have to do that | 18:23 |
| jroll | lol | 18:23 |
| jroll | I'll see if there's something we can do about that | 18:24 |
| jroll | idk what layer logs that | 18:24 |
| adam_g | jroll, devananda is this something we can build into ironicclient functional tests instead of tempest? that was what i was hoping to get started with https://review.openstack.org/#/c/161976/ | 18:24 |
| jroll | devananda: that's probably the best way, but alternatively ironic could log it | 18:24 |
| jroll | adam_g: I started a todo list for this stuff under gate status https://etherpad.openstack.org/p/IronicWhiteBoard | 18:25 |
| jroll | adam_g: oh wow, we have functional tests. you're awesome. | 18:25 |
| jroll | or a framework for them, at least | 18:25 |
| adam_g | jroll, well, we have somewhere to put them. | 18:25 |
| jroll | yeah | 18:25 |
| adam_g | i started looking at at least adding tests for version negotiating last week, but was going to wait for the api spec to finish up | 18:26 |
| devananda | adam_g: spec is landed now. but client is still broken | 18:26 |
| adam_g | on a similar topic, i just hit a virt bug in our partial upgrade story @ https://bugs.launchpad.net/ironic/+bug/1433727 that relates to API versioning | 18:26 |
| openstack | Launchpad bug 1433727 in Ironic "partial upgrade not possible, 'reason': u'Unknown argument: "configdrive" (HTTP 400)'" [Undecided,New] | 18:26 |
| * BadCub grumbles at Jenkins | 18:26 | |
| *** rameshg87 has left #openstack-ironic | 18:27 | |
| adam_g | devananda, broken in more ways that one, or just wrt version negotiation? | 18:27 |
| *** rameshg87 has joined #openstack-ironic | 18:28 | |
| adam_g | my assumption was we need the client functional test suite to be able to test and use negotiation so it can discover what versions of the api it should be testing | 18:28 |
| *** ramineni1 has quit IRC | 18:28 | |
| * rameshg87 goes to sleep | 18:28 | |
| rameshg87 | good night ironic | 18:28 |
| jlvillal | rameshg87: Good night! | 18:28 |
| NobodyCam | night rameshg87 | 18:28 |
| *** rameshg87 has left #openstack-ironic | 18:28 | |
| jroll | adam_g: on that bug... I thought we agreed that kilo upgrade process was upgrade ironic first, then nova. we had to make a one or the other choice for a few reasons | 18:29 |
| adam_g | jroll, thats for an full upgrade, this is a partial--where we're trying to make sure a kilo nova can spawn things on a juno ironic | 18:30 |
| adam_g | unless thats been decided to be not supported ? | 18:30 |
| jroll | adam_g: right, I think we decided that isn't supported. maybe I'm wrong. | 18:30 |
| jroll | my memory is crap right now | 18:30 |
| adam_g | oh | 18:30 |
| jroll | so don't take my word for that | 18:30 |
| jroll | devananda: ^^ | 18:30 |
| *** harmw has left #openstack-ironic | 18:30 | |
| *** kkoski has joined #openstack-ironic | 18:31 | |
| *** kkoski has quit IRC | 18:31 | |
| jroll | that said... we may be able to fix this | 18:31 |
| *** kkoski has joined #openstack-ironic | 18:31 | |
| jroll | well, maybe it takes proper client version negotiation | 18:32 |
| jroll | devananda: configdrive is another example of an api change between juno and having microversions | 18:32 |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: Automate uefi boot iso creation for iscsi_ilo driver https://review.openstack.org/155900 | 18:33 |
| *** kkoski has quit IRC | 18:33 | |
| adam_g | jroll, that was my thought. just fallback to not specifying config drive (or erroring?) if the drivers client can determine its not supported | 18:33 |
| jroll | adam_g: yeah, but the client isn't aware of what the server supports today :/ | 18:34 |
| devananda | adam_g: let's assume that juno nova is running with the juno release of python-ironicclient for a minute | 18:35 |
| devananda | adam_g: in that case, i would expeect juno nova to work with kilo ironic -- because the kilo ironic API will "downgrade" communication to match the older client | 18:36 |
| openstackgerrit | Merged openstack/ironic: Sync from oslo.incubator https://review.openstack.org/162505 | 18:36 |
| adam_g | devananda, right--but the partial upgrade job (as it is today) is the inverse | 18:36 |
| jroll | "juno release of python-ironicclient" | 18:36 |
| devananda | also, fwiw, that initial assumption is false, becuase nova refuses to add python-ironicclient to their requirements, so it wont be pinned | 18:36 |
| adam_g | devananda, oh, :| | 18:37 |
| devananda | so anyone installing stable/juno of nova from packages, or from pip, will still get the current version of python-ironicclient ... which is why i'm making so damn much fuss over this client change | 18:37 |
| devananda | because it WILL break that path if we're not really careful | 18:37 |
| devananda | adam_g: so the supported upgrade path is, as you say, the inverse. upgrade nova first, then upgrade ironic. | 18:37 |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: follow up patch for ilo capabilities https://review.openstack.org/165310 | 18:38 |
| devananda | jroll: right ^ ? | 18:38 |
| adam_g | devananda, *and python-ironicclient on the nova side | 18:38 |
| devananda | adam_g: yes | 18:38 |
| devananda | though seriously I think it shouldn't matter which order I upgrade nova or ironic in | 18:38 |
| adam_g | FWIW the experimental partial grenade job seems to pick up the newer client | 18:38 |
| adam_g | devananda, agreed | 18:38 |
| devananda | adam_g: yeah, i would expect that it picks up the newer client -- even in the "old" half of the grenade job | 18:39 |
| jroll | erm | 18:39 |
| * jroll thinks really hard | 18:39 | |
| devananda | adam_g: i mean, i expect that today. I think that's broken, however | 18:39 |
| devananda | adam_g: nova needs to pin the version of python-ironicclient on stable/juno. and on stable/kilo | 18:39 |
| devananda | *and on stable/kilo it should matter less ... assuming we sort all this out ... but it wouldn't hurt | 18:39 |
| devananda | jroll: oh. crap. | 18:41 |
| adam_g | i need to step away for 1hr but will take a look at the client functional stuff when im back | 18:41 |
| jroll | adam_g: devananda: alternatively, think about how a deployer is doing this upgrade. they read the docs and see configdrive is supported in kilo ironic. they don't start sending a configdrive while they're stillr unning juno ironic. | 18:41 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 18:41 |
| devananda | jroll: it's the nostate->available rename that concerns me | 18:41 |
| jroll | adam_g: devananda: IMO this is a config thing, kilo nova with juno ironic should be changed to not use configdrive because it isn't supported | 18:42 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for iscsi_ilo driver https://review.openstack.org/154814 | 18:42 |
| devananda | jroll: if i tag a client release that ups the api version and picks up that rename, it will immediately breat juno/nova | 18:42 |
| jroll | devananda: mmm. yeah. hm. | 18:43 |
| *** athomas has quit IRC | 18:43 | |
| devananda | jroll: so 1) i want the client to default to sending the latest version. 2) if it does that, it MUST also support downgrading to older versions (which we have absolutely no code for today) 3) because juno/nova isn't pinned, if I do either of those, we break juno/nova | 18:43 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for agent_ilo driver https://review.openstack.org/154816 | 18:43 |
| devananda | therefor: our client has to continue defaulting to the lowest version | 18:43 |
| devananda | *bangs head on desk* | 18:44 |
| jroll | devananda: indeed. I think we should fix (2) real hard real soon, because that's a huge UI pain point too | 18:44 |
| *** athomas has joined #openstack-ironic | 18:44 | |
| devananda | jroll: https://review.openstack.org/#/c/162834/1 | 18:45 |
| devananda | i started on that last week | 18:45 |
| *** kkoski has joined #openstack-ironic | 18:46 | |
| devananda | but i only did it in the shell so far, not the client lib | 18:46 |
| *** dprince has quit IRC | 18:46 | |
| devananda | oh bugger. i lost a change there :( | 18:46 |
| jroll | devananda: cool | 18:46 |
| devananda | hmmm. how do I list orphaned commits? | 18:47 |
| *** pelix has quit IRC | 18:47 | |
| NobodyCam | brb | 18:47 |
| *** kkoski has quit IRC | 18:47 | |
| devananda | found it | 18:48 |
| *** achanda has quit IRC | 18:48 | |
| *** athomas has quit IRC | 18:48 | |
| openstackgerrit | Merged openstack/ironic: iLO implementation for hardware inspection https://review.openstack.org/151596 | 18:49 |
| openstackgerrit | Merged openstack/ironic: iLO driver updates node capabilities during inspection https://review.openstack.org/163572 | 18:49 |
| jlvillal | devananda: reflog? | 18:49 |
| jlvillal | nvm | 18:49 |
| *** hj-hp has quit IRC | 18:51 | |
| *** Marga_ has quit IRC | 18:51 | |
| openstackgerrit | Devananda van der Veen proposed openstack/python-ironicclient: Send version header by default https://review.openstack.org/165559 | 18:51 |
| openstackgerrit | Devananda van der Veen proposed openstack/python-ironicclient: Hide resources missing from server responses - POC - DO NOT MERGE https://review.openstack.org/162834 | 18:52 |
| *** Marga_ has joined #openstack-ironic | 18:53 | |
| openstackgerrit | Devananda van der Veen proposed openstack/python-ironicclient: Adjust CLI output based on API version https://review.openstack.org/162834 | 18:54 |
| *** stendulker has joined #openstack-ironic | 18:58 | |
| *** kkoski has joined #openstack-ironic | 18:58 | |
| *** kkoski has quit IRC | 18:58 | |
| *** kkoski has joined #openstack-ironic | 18:59 | |
| *** dprince has joined #openstack-ironic | 19:00 | |
| *** kkoski has quit IRC | 19:01 | |
| *** achanda has joined #openstack-ironic | 19:01 | |
| *** hj-hp has joined #openstack-ironic | 19:03 | |
| *** alexpilotti_ has quit IRC | 19:04 | |
| * NobodyCam is back | 19:04 | |
| *** Marga_ has quit IRC | 19:04 | |
| *** Marga_ has joined #openstack-ironic | 19:05 | |
| *** andreykurilin_ has joined #openstack-ironic | 19:06 | |
| *** jiangfei|2 has joined #openstack-ironic | 19:06 | |
| *** Nisha has quit IRC | 19:07 | |
| NobodyCam | Nisha: are you still about? | 19:07 |
| *** jiangfei has quit IRC | 19:08 | |
| openstackgerrit | Merged openstack/ironic: iscsi_ilo driver do not validate boot_option https://review.openstack.org/164414 | 19:09 |
| NobodyCam | I'm looking at 155900 images.py, on line 289 yuriyz commented on moving some writes to try block. looks like you answered ok. but I don't see the changes. | 19:09 |
| *** kkoski has joined #openstack-ironic | 19:12 | |
| openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add fallback secure erase using shred https://review.openstack.org/165247 | 19:14 |
| JayF | JoshNang: ^ fixed your comments, improved the shred output in the test, and added an additional test | 19:15 |
| *** kkoski has quit IRC | 19:15 | |
| * JoshNang reviews | 19:15 | |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 19:15 |
| JoshNang | JayF: sweet, lgtm | 19:18 |
| stendulker | NobodyCam: Hi :) | 19:19 |
| stendulker | NobodyCam: This is regarding your comment for review https://review.openstack.org/#/c/154814/ Secure boot support for iscsi_ilo driver | 19:20 |
| *** hj-hp has quit IRC | 19:21 | |
| JoshNang | devananda: as i'm fixing up the agent cleaning patch, also noticed prepare_cleaning (which boots the ramdisk) also uses the rpc call back. i'm not sure where to go from here. | 19:21 |
| jroll | JoshNang: I think connecting the task to the manager might be the right thing to do here | 19:22 |
| *** kkoski has joined #openstack-ironic | 19:22 | |
| stendulker | NobodyCam: Have responded to your comment. I need to re-post the patch to resolve merge conflicts. In case anything required to address your comment I can add it as well. If you have a moment you can check the comment resolution. | 19:22 |
| devananda | jroll: i looked briefly at that. it's not trivial :( | 19:22 |
| JoshNang | :/ | 19:22 |
| jroll | devananda: optional kwarg for task.acquire? | 19:22 |
| NobodyCam | stendulker: let me have a quick look | 19:22 |
| stendulker | NobodyCam: I can always repost the patch later as well. But just wanted to check I could accomodate | 19:22 |
| JoshNang | that was my thought | 19:23 |
| *** romcheg has quit IRC | 19:23 | |
| JoshNang | we'd only need the reference in a few places (to start) | 19:23 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Ilo drivers sets capabilities:boot_mode in node https://review.openstack.org/155731 | 19:23 |
| devananda | JoshNang: yah, maybe | 19:23 |
| NobodyCam | stendulker: Thank you, can you the "rational" bit to the commit message when you re-post! that will cover it for me! | 19:25 |
| NobodyCam | *can you MOVE the ... | 19:25 |
| NobodyCam | s/MOVE/ADD. | 19:25 |
| TheJulia | JayF: fyi, A kernel param of coreos.configdrive=0 prevents the config drive from being auto-mounted. | 19:25 |
| *** hj-hp has joined #openstack-ironic | 19:25 | |
| stendulker | NobodyCam: Ok. Will do that. | 19:26 |
| NobodyCam | awesome TY stendulker :) | 19:26 |
| stendulker | NobodyCam: Thank you for quickly validating the resolution :) | 19:26 |
| *** andreykurilin_ has quit IRC | 19:27 | |
| *** romcheg has joined #openstack-ironic | 19:28 | |
| * devananda adds links to client patches to ehterpad | 19:28 | |
| devananda | JoshNang: shall i tackle the task manager change? | 19:28 |
| devananda | JoshNang: or are you confident that using the rpc callback from the agent driver is not going to break the world? | 19:29 |
| *** Nisha has joined #openstack-ironic | 19:30 | |
| * BadCub needs food like thing | 19:31 | |
| NobodyCam | oh like manna! :-p | 19:32 |
| BadCub | less mythical than manna :-p | 19:33 |
| jroll | devananda: I plan to run the RC downstream in the next couple weeks | 19:33 |
| jroll | fwiw | 19:33 |
| jroll | so we can at least find out if it scales | 19:34 |
| rloo | devananda: sorry to bug you. Just noticed that https://review.openstack.org/#/c/163730/ was bumped to liberty? | 19:34 |
| rloo | devananda: can't that be done before rc-1/kilo release? | 19:34 |
| *** aswadr has quit IRC | 19:34 | |
| JoshNang | devananda: either works. i've only tested it with 3 nodes at the same time | 19:35 |
| JoshNang | anyone got a sec to fix a merge conflict on an approved patch? https://review.openstack.org/#/c/157715/ | 19:35 |
| devananda | Nisha: your commit 6ecee368 added inspect*at fields to the API, but did not use any microversion to guard them :-( | 19:36 |
| devananda | Nisha: so those are now being returned with the base API version | 19:36 |
| rloo | devananda: forget I asked about 163730. There's a string change there so it can't go in after k-3. | 19:37 |
| devananda | dtantsur|afk: this sort of things kind of goes to your point ^^ -- on March 2nd, we landed a change to the API with no version guads | 19:37 |
| devananda | rloo: ack. forgetting :) | 19:38 |
| devananda | jroll: good to know. while i appreciate tht, i also dont want to find out that late (and break ya'll inthe process) | 19:38 |
| devananda | JoshNang: 3 deployed nodes, or 3 conductors? | 19:39 |
| *** Ng has quit IRC | 19:39 | |
| Nisha | devananda, :( | 19:39 |
| JoshNang | devananda: 3 deployed nodes. only 1 conductor (devstack) | 19:39 |
| *** krtaylor has quit IRC | 19:39 | |
| *** comstud has quit IRC | 19:39 | |
| *** rainya has quit IRC | 19:39 | |
| *** Ng has joined #openstack-ironic | 19:39 | |
| devananda | JoshNang: it's easy to create more conductors in devstack -- copy /etc/ironic/ironic.conf, add "hostname=NNNN", and start another one in screen | 19:40 |
| devananda | JoshNang: nodes will remap accordingly as the ring rebalances | 19:40 |
| devananda | *start another one and specify --config-file=/path/to/new/file | 19:40 |
| JoshNang | ok i'll test | 19:40 |
| *** rainya has joined #openstack-ironic | 19:40 | |
| *** comstud has joined #openstack-ironic | 19:41 | |
| jroll | devananda: we have around 50 nodes in a preprod environment, think that would be enough scale to find out? | 19:41 |
| devananda | JoshNang: and while that's going on, take a look at rabbit | 19:41 |
| devananda | jroll: yup | 19:41 |
| devananda | TheJulia: also you might be able to help test this, if you want to try pulling a bunch of patches from gerrit? | 19:41 |
| *** krtaylor has joined #openstack-ironic | 19:41 | |
| jroll | yeah, so I could get this running within a couple days of it landing | 19:42 |
| *** kkoski has quit IRC | 19:42 | |
| *** Marga_ has quit IRC | 19:43 | |
| *** kkoski has joined #openstack-ironic | 19:43 | |
| JoshNang | yeah i think we were planning on staging this next week | 19:43 |
| *** Marga_ has joined #openstack-ironic | 19:44 | |
| devananda | JoshNang: is there an easy way for me to simulate this // mock the agent so that the agent driver does the rpc call back without actually cleaning a node? | 19:44 |
| JoshNang | devananda: set the config option to 0. sec.. | 19:44 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for iscsi_ilo driver https://review.openstack.org/154814 | 19:45 |
| JoshNang | err, that'll still try to call out to the agent to get the clean steps | 19:45 |
| *** wanyen has quit IRC | 19:46 | |
| JoshNang | devananda: return [] from get clean_steps in the agent? | 19:47 |
| JoshNang | *agent driver | 19:47 |
| devananda | Nisha: since you landed the inspect*at base patch, can you post a fix that properly guards that with a version check and bump? | 19:49 |
| Nisha | Yes i am working on that | 19:50 |
| Nisha | trying to post it by today | 19:50 |
| *** ijw has quit IRC | 19:53 | |
| *** Marga_ has quit IRC | 19:54 | |
| JayF | TheJulia: if you look at the underlying unit files, that's what masking all those units is supposed to do | 19:54 |
| JayF | TheJulia: So I'm still perplexed as to why that happened? | 19:55 |
| *** devlaps has joined #openstack-ironic | 19:55 | |
| *** zigo__ has joined #openstack-ironic | 19:56 | |
| *** zigo has quit IRC | 19:56 | |
| devananda | Nisha: thanks much | 19:57 |
| *** rainya has quit IRC | 19:57 | |
| * JoshNang kicks off 3 cleans, 3 conductors | 19:57 | |
| devananda | JoshNang: watch rabbit queues | 19:57 |
| *** rainya has joined #openstack-ironic | 19:58 | |
| *** subscope_ has joined #openstack-ironic | 19:58 | |
| devananda | JoshNang: actually not just queues. watch list_consumers too | 19:59 |
| JoshNang | kk | 19:59 |
| TheJulia | JayF: the mount still seems to get triggered even if the task is masked, so at least it doesn't auto-mount anymore when I use the kernel command line:) | 19:59 |
| * devananda foods and meetings | 19:59 | |
| JayF | TheJulia: If you can give me full logs from a boot where it mounted in into a bug, I can track it down | 20:00 |
| *** achanda has quit IRC | 20:00 | |
| NobodyCam | devananda: enjoy! | 20:00 |
| *** ifarkas has quit IRC | 20:01 | |
| BadCub | jroll: can I rebase 157715 to clear the merge conflict? Or does it need to be done manually> | 20:01 |
| jroll | BadCub: if there's a conflict it needs to be done manually | 20:02 |
| jroll | if it could do it automatically it would ;) | 20:02 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for agent_ilo driver https://review.openstack.org/154816 | 20:03 |
| * NobodyCam thinks BadCub just wants to click the rebase button :-p | 20:03 | |
| * BadCub stares at the shiny button | 20:03 | |
| NobodyCam | https://www.youtube.com/watch?v=FT4YbO_1mvA | 20:05 |
| NobodyCam | lol | 20:05 |
| jroll | I mean, click away but it won't help :P | 20:06 |
| *** dprince has quit IRC | 20:06 | |
| BadCub | NobodyCam: LOL | 20:06 |
| jroll | lol | 20:06 |
| NobodyCam | lol I love the laughing at the end | 20:07 |
| BadCub | DogTV music is killing me... Must change channel | 20:07 |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: Automate uefi boot iso creation for iscsi_ilo driver https://review.openstack.org/155900 | 20:08 |
| NobodyCam | can rebase 157715 | 20:09 |
| JoshNang | devananda: all the nodes did continue_node_cleaning, all queues (except notification.info) are sitting at 0, consumers is 3 or less everywhere | 20:10 |
| JoshNang | (still waiting for cleaning to finish, as shred take...a while :/) | 20:10 |
| TheJulia | JayF: Will do, still trying to get my configdrive written out :) | 20:13 |
| *** openstackgerrit has quit IRC | 20:14 | |
| *** dprince has joined #openstack-ironic | 20:15 | |
| *** openstackgerrit has joined #openstack-ironic | 20:15 | |
| openstackgerrit | Merged openstack/ironic: Suppress urllib3.connection INFO level logging https://review.openstack.org/165455 | 20:15 |
| JoshNang | *cough* running cleaning in devstack with shredding enabled is not advisable. thinly provisioned disks + shredding = full disk | 20:15 |
| JoshNang | (forgot to shrink disks) | 20:16 |
| *** jiangfei|3 has joined #openstack-ironic | 20:16 | |
| *** jiangfei|2 has quit IRC | 20:16 | |
| openstackgerrit | Chris Krelle proposed openstack/ironic: Add Cleaning Operations for iLO drivers https://review.openstack.org/157715 | 20:18 |
| NobodyCam | JoshNang: BadCub: rebased ^^^^ | 20:18 |
| JoshNang | NobodyCam: thanks! | 20:18 |
| BadCub | thanks NobodyCam :-) | 20:20 |
| devananda | JoshNang: how many queues and how many consumers are there total? | 20:20 |
| *** Marga_ has joined #openstack-ironic | 20:21 | |
| JoshNang | devananda: umm i had to unstack/restack. | 20:22 |
| devananda | JoshNang: oh | 20:22 |
| JoshNang | (filled up the disks by shredding 3 nodes without resizing their disks first) | 20:22 |
| devananda | hehe | 20:23 |
| NobodyCam | GheRivero: are you around? | 20:23 |
| JayF | I feel like I should warn our cloud ops team that ironic test nodes may be a lot less friendly of an IO neighbor real soon now | 20:23 |
| JayF | lol | 20:23 |
| *** hj-hp has quit IRC | 20:24 | |
| GheRivero | NobodyCam: in a call, but tell me | 20:24 |
| jroll | JayF: lol | 20:24 |
| NobodyCam | hehehe : can you dbl check and see if this change looks sane to you: https://review.openstack.org/#/c/155900/13/etc/ironic/ironic.conf.sample | 20:24 |
| GheRivero | NobodyCam: added to my list for tonight | 20:26 |
| NobodyCam | :) | 20:26 |
| NobodyCam | ty GheRivero :) | 20:26 |
| devananda | JayF: you did the fallback to full erase, i take it. should we do a "if youre in a vm, skip" step as well? | 20:28 |
| JayF | devananda: That's /really hard/ to detect without invalidating all of our testing. | 20:29 |
| devananda | JayF: you mentioned it yesterday, but yea, that might be a big surprise to anyone testing with this (including our upstream gate) | 20:29 |
| JayF | devananda: This is why I didn't want to write that patch at all, hehe | 20:29 |
| devananda | ah | 20:29 |
| JayF | It's almost like I want to use a devstack hardware manager | 20:29 |
| JayF | that just walks through some fake steps | 20:29 |
| JayF | since really the coordination is what we're testing, not the viability of the actual cleaning step being performced | 20:30 |
| JayF | *performed | 20:30 |
| devananda | JayF: right | 20:30 |
| devananda | JayF: so if we DONT dothat, how much longer will our gate tests take? | 20:30 |
| JayF | but I'm not really sure how to go about that detection ... it honestly scares me a little trying to detect, b/c if we fail, that's a security vuln in whoevers environment we didn't clean | 20:30 |
| JayF | devananda: I know shredding a 29GB disk took about 20 minutes | 20:30 |
| JoshNang | yeahhh it's quite a while | 20:31 |
| JoshNang | we can shrink the disks to save some time | 20:31 |
| devananda | :-/ | 20:31 |
| JoshNang | or, i added a config to skip it..but then we're not really testing cleaning. | 20:31 |
| JoshNang | less we add some fake steps or something | 20:31 |
| devananda | or we end up with separate IPA ramdisks for gate // prod | 20:32 |
| devananda | which is also less than ideal | 20:32 |
| JayF | Or maybe you end up implementing the "skip categories of steps" thing for the agent | 20:32 |
| Shrews | or, our drivers could have loadable cleaning drivers and we create a fake one | 20:32 |
| * Shrews hides | 20:32 | |
| JayF | Shrews: HardwareManager is the word you're looking for, and it's what I proposed above ^ | 20:32 |
| devananda | like we pass some parameters to the agent to say "run these things" | 20:32 |
| JayF | :( | 20:33 |
| JayF | devananda: explicitly not | 20:33 |
| devananda | and we have a seprate gate job that tests "do all the hardware stuff, we know it'll be slow" | 20:33 |
| devananda | and the main job skips clenning, or something ? | 20:33 |
| JayF | devananda: it's *very important* that the agent determine the steps it runs for hardware detection support | 20:33 |
| devananda | JayF: categories. not individual hardware anager stes | 20:33 |
| devananda | steps | 20:33 |
| JayF | I'm OK with that | 20:33 |
| devananda | so gate conig would be "skip disk erase" right now inthe main job | 20:33 |
| devananda | or something ... | 20:34 |
| devananda | *shrug* | 20:34 |
| JoshNang | yeah i think that'd work | 20:34 |
| JayF | Yeah. | 20:34 |
| JayF | JoshNang: you gonna write that? | 20:34 |
| JayF | JoshNang: we also did a *wonderful job* of talking about cleaning docs this morning | 20:34 |
| JayF | heh | 20:34 |
| JoshNang | JayF: heh | 20:35 |
| JoshNang | well i have the config option to disable disk erase already | 20:35 |
| *** Haomeng|2 has joined #openstack-ironic | 20:37 | |
| *** Haomeng has quit IRC | 20:40 | |
| jroll | JayF: devananda: you could have devstack pull in the out-of-tree devstack hardware manager | 20:40 |
| jroll | that way you don't have to worry much about detection | 20:41 |
| devananda | jroll: into the ipa ramdisk? | 20:41 |
| JayF | jroll: I thought about that, but that sketched me out a little ... I don't want us to test things that aren't Ironic/IPA ... I'd prefer the "disable stuff via config" and just flip those configs on in the gate | 20:41 |
| devananda | JayF: ++ | 20:41 |
| jroll | oh. it would need to build the ramdisk | 20:41 |
| jroll | lame | 20:41 |
| devananda | right | 20:42 |
| adam_g | FWIW cinder skips secure delete in the gate for the same reason | 20:42 |
| adam_g | devananda, catching up on scrollback (sorry): "so 1) i want the client to default to sending the latest version. 2) if it does that, it MUST also support downgrading to older versions (which we have absolutely no code for today) 3) because juno/nova isn't pinned, if I do either of those, we break juno/nova" <- how would adding that negotiating break juno? | 20:45 |
| adam_g | assuming something similar to https://review.openstack.org/#/c/162834/4/ironicclient/v1/client.py ln 56 | 20:45 |
| *** mrda-away is now known as mrda | 20:47 | |
| mrda | Morning Ironic | 20:47 |
| mrda | devananda: thanks for the pickup | 20:47 |
| BadCub | mrda: morning | 20:47 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 20:48 |
| *** achanda has joined #openstack-ironic | 20:48 | |
| mrda | Morning BadCub | 20:48 |
| NobodyCam | mornign mrda :) | 20:49 |
| mrda | o/ | 20:50 |
| * NobodyCam steps away for a few minutes | 20:50 | |
| devananda | adam_g: juno/nova's ironic driver only understands specific node states (eg, NOSTATE) | 20:51 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for pxe_ilo driver https://review.openstack.org/154808 | 20:51 |
| devananda | adam_g: if the client library uses any version above 1.1 to talk with a kilo ironic server, it's going to return states to nova.virt.ironic which tht code does not understand | 20:51 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for iscsi_ilo driver https://review.openstack.org/154814 | 20:52 |
| JayF | devananda: I thought clients got frozen to stable branches now? | 20:52 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Secure boot support for agent_ilo driver https://review.openstack.org/154816 | 20:52 |
| devananda | adam_g: in other words, the stable/juno branch of nova needs to be pinned to python-ironicclient < 0.5 | 20:52 |
| JayF | devananda: so couldn't we have a client that nova used for k->j that did it one way, then move forward with the new way? | 20:52 |
| JayF | yeah, exactly | 20:52 |
| devananda | JayF: they do. but nova refuses to put pyhton-ironicclien tin their requirements | 20:52 |
| openstackgerrit | Shivanand Tendulker proposed openstack/ironic: Ilo drivers sets capabilities:boot_mode in node https://review.openstack.org/155731 | 20:52 |
| devananda | therefor it can't be frozen | 20:52 |
| *** ParsectiX has quit IRC | 20:52 | |
| mrda | oh, that's right | 20:53 |
| adam_g | devananda, what about a patch to stable/juno nova to specify v1.1 to the client? | 20:53 |
| *** ParsectiX has joined #openstack-ironic | 20:53 | |
| devananda | JayF: right. that's what i'm suggesting. the current client works -- because it doesnt send any header | 20:53 |
| mrda | dynamic loading as a result | 20:53 |
| devananda | adam_g: ++ | 20:53 |
| JayF | ==adam_g | 20:53 |
| adam_g | is there a bug to reference somewhere? | 20:53 |
| devananda | nope | 20:54 |
| adam_g | k | 20:54 |
| devananda | no bug because we hvaen't actually broken anything yet | 20:54 |
| devananda | heh | 20:54 |
| *** hj-hp has joined #openstack-ironic | 20:54 | |
| devananda | filing a bug before we break something is a bit .... odd :) | 20:54 |
| adam_g | no way! | 20:55 |
| *** alexpilotti has joined #openstack-ironic | 20:56 | |
| openstackgerrit | Merged openstack/ironic: Allow periods (".") in hostnames https://review.openstack.org/165443 | 20:58 |
| *** stendulker has quit IRC | 20:59 | |
| jlvillal | devananda: Not sure if this is a dumb idea. Would there be any value in the python-ironicclient querying the nova server to figure out it's version. If that is possible. Not sure if that would be useful or not. | 21:01 |
| devananda | jlvillal: it's a library, loaded by the nova.virt.ironic code | 21:01 |
| jlvillal | You have probably already considered that idea, but thought I'd mention it. | 21:01 |
| devananda | jlvillal: we can't change the nova code that someone is already running in production // has been packaged by distros | 21:02 |
| devananda | but if we change the library, it could get sucked into their environment (and it will, in the gate, if it's not pinned) | 21:02 |
| openstackgerrit | Ruby Loo proposed openstack/ironic: Hide inspection_*_at fields if version < 1.6 https://review.openstack.org/165596 | 21:02 |
| *** jiangfei|3 has quit IRC | 21:02 | |
| jlvillal | devananda: I was thinking maybe change what the ironicclient does based on determining the nova version. But maybe not feasible. | 21:02 |
| *** jiangfei|3 has joined #openstack-ironic | 21:03 | |
| * jlvillal admits he doesn't fully understand how it all works together. | 21:03 | |
| *** kkoski has quit IRC | 21:03 | |
| devananda | jlvillal: the library can't reasonably determine the version of the thing that calls it | 21:05 |
| devananda | without doing some funky dance inspecting the python call stack .... | 21:05 |
| jlvillal | devananda: Okay. Never mind then :) Thanks for answering. | 21:05 |
| devananda | jlvillal: it's a neat idea -- but i dont think it's possible, unfortunately | 21:06 |
| *** jiangfei|3 has quit IRC | 21:09 | |
| *** jiangfei|3 has joined #openstack-ironic | 21:09 | |
| *** lucas-dinner has quit IRC | 21:10 | |
| adam_g | devananda, jroll is this accurate? https://etherpad.openstack.org/p/ir-nova-client would like to make sure its accurate before filing it, i expect it might be contentious on the nova side | 21:12 |
| *** kkoski has joined #openstack-ironic | 21:13 | |
| *** dprince has quit IRC | 21:14 | |
| mrda | adam_g: I think that's fair. | 21:14 |
| adam_g | mrda, thanks | 21:16 |
| BadCub | looks good to me adam_g :-) | 21:16 |
| rloo | mrda: hi. fyi, in case you didn't see it: https://review.openstack.org/#/c/165443/ | 21:17 |
| *** kkoski has quit IRC | 21:17 | |
| *** ChuckC has quit IRC | 21:19 | |
| * NobodyCam is back | 21:20 | |
| NobodyCam | jlvillal: here is how it all works: https://i.imgur.com/1CXQNjw.png | 21:20 |
| mrda | thanks rloo | 21:21 |
| BadCub | lol | 21:21 |
| *** ParsectiX has quit IRC | 21:21 | |
| *** ParsectiX has joined #openstack-ironic | 21:21 | |
| mrda | rloo: I'll follow up with another patch, because that is an incorrect implementation unfortunately | 21:22 |
| rloo | mrda: ha ha. thx! | 21:22 |
| mrda | atom = '^[a-z0-9]([a-z0-9\-\.]{0,61}[a-z0-9])?$' | 21:23 |
| mrda | match_str = atom(\.atom)+ | 21:23 |
| mrda | (cut'n'paste error, atom shouldn' | 21:23 |
| mrda | t have a . in it) | 21:24 |
| jlvillal | NobodyCam: Thanks! | 21:24 |
| NobodyCam | :-p | 21:24 |
| mrda | It's actually more complex than that - I'll find a regex that fully implements the rfc | 21:24 |
| rloo | mrda: good catch. (I didn't actually look at the patch. I have a love/hate relationship with regexs.) | 21:25 |
| mrda | You can have _'s in the domain part, but not the host part (IIRC). I'll have to go check RFCs again. | 21:26 |
| mrda | Just goes to show that being away from ironic for 6 hours means you miss patches getting proposed and merged :) | 21:26 |
| *** jmccrory has joined #openstack-ironic | 21:27 | |
| rloo | mrda: "we" are fast and trigger-happy :-) | 21:28 |
| jlvillal | NobodyCam: Not quite as useful as I was expecting :P | 21:28 |
| jlvillal | LOL | 21:28 |
| mrda | I should also push a patch to get the spec updated, because having broken documentation deliberately isn't cool | 21:28 |
| NobodyCam | lol oh you just looked :-p | 21:28 |
| NobodyCam | hehehe | 21:28 |
| rloo | mrda: thx. I was hoping you'd offer to do that! | 21:28 |
| mrda | "should" isn't a commitment, and I have higher priority things to do too | 21:29 |
| mrda | but I probably will | 21:29 |
| mrda | post-coffee | 21:29 |
| *** jamielennox is now known as jamielennox|away | 21:31 | |
| mrda | Gack. 165443 is broken in another way. The total length of the hostname(.domain)+ is 63 chars right now. To fix this properly needs another db migration. | 21:35 |
| mrda | each element of hostname/domain should be up to 63 chars, not the total length. | 21:36 |
| mrda | sigh. | 21:36 |
| NobodyCam | yep for a total of 255 char | 21:37 |
| NobodyCam | le sigh | 21:37 |
| *** ParsectiX has quit IRC | 21:38 | |
| *** ParsectiX has joined #openstack-ironic | 21:38 | |
| rloo | mrda: good, better now than in L*. (Well, that's the bright side anyway.) | 21:38 |
| *** Marga_ has quit IRC | 21:39 | |
| * mrda doesn't like changes rushed through | 21:40 | |
| *** Marga_ has joined #openstack-ironic | 21:40 | |
| *** Marga_ has quit IRC | 21:40 | |
| JoshNang | devananda: queues/consumers (with a link at the bottom to during cleaning) https://gist.github.com/pcsforeducation/6558b00c2b1683df6503 | 21:40 |
| JoshNang | i don't see anything off | 21:40 |
| * BadCub steps away for a few minutes | 21:40 | |
| *** Marga_ has joined #openstack-ironic | 21:40 | |
| JoshNang | but i don't look at this very often | 21:41 |
| devananda | adam_g: commentes added on the 'pad | 21:43 |
| *** devlaps has quit IRC | 21:43 | |
| adam_g | devananda, thanks | 21:43 |
| devananda | JoshNang: i haven't debugged this stuff since we wrote teh rpc api, so yea, not immediately familiar with it.. but ... | 21:44 |
| *** hj-hp has quit IRC | 21:45 | |
| devananda | JoshNang: yea, not immediately scary. I'm unsure as to why there are 9 "reply" queues though | 21:47 |
| *** hj-hp has joined #openstack-ironic | 21:47 | |
| JoshNang | that is odd...though they were there before i started running anything | 21:48 |
| JoshNang | (though 2 more added during cleaning. gonna run again and see if those keep growing) | 21:49 |
| *** ParsectiX has quit IRC | 21:49 | |
| *** persia has quit IRC | 21:50 | |
| *** pcrews has quit IRC | 21:51 | |
| *** persia has joined #openstack-ironic | 21:51 | |
| *** persia has quit IRC | 21:51 | |
| *** persia has joined #openstack-ironic | 21:51 | |
| *** pcrews has joined #openstack-ironic | 21:51 | |
| * devananda makes toast with nutella and peanutbutter | 21:52 | |
| NobodyCam | mmmmm | 21:52 |
| NobodyCam | :-p | 21:52 |
| JayF | jlvillal: re: your review on my IPA patch, utils.execute will log a bunch of stuff on failure, I don't think IPA needs a log too, but I can lower the scope of that except | 21:52 |
| * NobodyCam has been having walffles all week | 21:52 | |
| *** kkoski has joined #openstack-ironic | 21:52 | |
| jlvillal | JayF: Fair enough. I just worry in general about except Exception. Since it catches so much. | 21:53 |
| jlvillal | JayF: Thanks for the info! | 21:53 |
| JayF | I mean, you're right-ish | 21:53 |
| JayF | there I didn't worry so much b/c I knew it'd end up with an errors.IncompatibleHardwareMethodError | 21:53 |
| JayF | but I should make the method behave reasonably on it's own | 21:53 |
| JayF | s/it's/its/ | 21:53 |
| openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent: Add fallback secure erase using shred https://review.openstack.org/165247 | 21:55 |
| JayF | JoshNang: jlvillal ^ fixed jlvillal's concern | 21:55 |
| devananda | jlvillal: i worry about those too | 21:55 |
| JoshNang | devananda: i'm not seeing the reply queues growing at all, though this silly 8gb vm won't let me go over 3 nodes | 21:56 |
| JoshNang | JayF: ++ | 21:57 |
| JoshNang | (also, cleaning goes much, much faster with 2gb disks) | 21:57 |
| *** kkoski has quit IRC | 21:59 | |
| jlvillal | JayF: Just curious. Is anything done with mocked_execute.side_effect. I know it was already there. I was just curious if that is used for anything. | 22:00 |
| JayF | jlvillal: yeah, that's what's returned when that thing is called, in order | 22:00 |
| JayF | jlvillal: so the first time utils.execute is called, it gets the hdparm output. the second time it gets the shred output. | 22:00 |
| JayF | jlvillal: we don't do anything with the shred output, so providing anything at all there is a bonus, but I'd prefer it be represenative of the actual response even if we don't use it | 22:01 |
| jlvillal | JayF: Ah okay. The last part makes it make sense :) Thanks. | 22:01 |
| jlvillal | Anyone else going to the PDX OpenStack Hackathon next week? | 22:02 |
| JayF | morgabra: ^ hey, there's a PDX openstack hackathon | 22:02 |
| * jlvillal thinks he is the only person doing Ironic stuff who is in the Portland area :) | 22:02 | |
| JayF | jlvillal: morgabra is on the OnMetal team and is in your general vicinity | 22:02 |
| jlvillal | JayF: Cool | 22:03 |
| JayF | jlvillal: although he stays locked in mortal combat with Neutron and the switches it controls at almost all times | 22:03 |
| jlvillal | morgabra: http://calagator.org/events/1250467647 I think it is being organized by an HP guy. | 22:03 |
| JayF | lol | 22:03 |
| jlvillal | morgabra: Spencer Krum | 22:03 |
| openstackgerrit | Chris Krelle proposed openstack/ironic: Check temp dir is usable for ipmitool driver https://review.openstack.org/160383 | 22:04 |
| morgabra | JayF: jlvillal: aww, won't be in town unfortunately | 22:04 |
| jlvillal | JayF: And thanks | 22:05 |
| morgabra | PSU is nice | 22:05 |
| JayF | Ah, I took the streetcar through that area when I went to PDX for Open Source Bridge | 22:05 |
| JayF | it was super nice | 22:05 |
| jlvillal | NobodyCam: +1 from me on patch :) | 22:07 |
| NobodyCam | Awesome :) Ty jlvillal :) | 22:07 |
| jlvillal | NobodyCam: Thanks for the patch :) | 22:07 |
| jlvillal | NobodyCam: You've got the hard part | 22:08 |
| NobodyCam | :) thank for your work on the same bug!!!! | 22:08 |
| jlvillal | Glad to help :) | 22:08 |
| NobodyCam | :) | 22:08 |
| *** kkoski has joined #openstack-ironic | 22:12 | |
| *** jamielennox|away is now known as jamielennox | 22:12 | |
| rfchapman | NobodyCam: If i need to set the serial port or other kernel command line additions, do I do that in ironic extra? | 22:14 |
| JayF | There's a template you can edit | 22:14 |
| JayF | to modify that, I think there's actually a different one for agent ramdisk vs bash ramdisk | 22:14 |
| NobodyCam | rfchapman: which driver pxe? | 22:15 |
| NobodyCam | https://github.com/openstack/ironic/blob/master/etc/ironic/ironic.conf.sample#L346 | 22:15 |
| rfchapman | ipmi | 22:16 |
| NobodyCam | ^^^ is for agent params | 22:16 |
| rfchapman | But what if I want to do it by node. | 22:16 |
| NobodyCam | and https://github.com/openstack/ironic/blob/master/etc/ironic/ironic.conf.sample#L1296 | 22:16 |
| rfchapman | pxe_ipmitool | 22:16 |
| NobodyCam | rfchapman: come back next cycle | 22:16 |
| NobodyCam | :-p | 22:17 |
| devananda | rfchapman: what doyou need? | 22:17 |
| NobodyCam | can we append kernel param per node | 22:17 |
| NobodyCam | the two link I just linked will append to all nodes | 22:17 |
| NobodyCam | not per node | 22:17 |
| GheRivero | NobodyCam: I just take a look to the patch. The config file is a bit outdated, but nothing important | 22:17 |
| devananda | pxe kernel parameter? no. | 22:17 |
| devananda | change the ipmi connection port? yes | 22:18 |
| NobodyCam | rfchapman: did I miss understand the question? your looking to pass kernel params? | 22:18 |
| NobodyCam | GheRivero: :) that was from a grnerate sample conf run today | 22:19 |
| rfchapman | I need to set the serial port speed and port on the kernal append line. How can that be done per node? | 22:20 |
| rfchapman | NobodyCam: yes | 22:20 |
| devananda | rfchapman: i'm confused. set serial port speed during boot ? | 22:21 |
| JayF | rfchapman: devananda: Yeah, it's a sensible question | 22:22 |
| JayF | rfchapman: do you not have a single serial console config you can use on all nodes? | 22:22 |
| rfchapman | Sorry console port and speed "console=ttyS1,115200" | 22:22 |
| NobodyCam | rfchapman: and that changes per node? | 22:23 |
| NobodyCam | if not just add to : https://github.com/openstack/ironic/blob/master/etc/ironic/ironic.conf.sample#L1296 | 22:23 |
| rfchapman | JayF: No I have a diverse collection of HP hardware. | 22:24 |
| JayF | Alternatively, if you aren't using the iscsi driver and are booting from disk once provisioned, you can use the "None" DHCP provider and run your own DHCP server ... in which case you can PXE with whatever options you want, but Ironic won't do it for you anymore :) | 22:24 |
| rfchapman | So there is no way to pass kernel params at boot per node? | 22:25 |
| NobodyCam | is this something we would want spec for in L? | 22:26 |
| * NobodyCam is starting to see node.extra['kernel_param_append'] when he closes his eyes | 22:27 | |
| rfchapman | I would hope I could customize the boot with some granularity... Some systems use ttyS0 and other ttyS1 some run an 9600 and other 115200 | 22:27 |
| JayF | rfchapman: Yeah, it's definately something we could support and I could envision lots of people needing it. That being said we're well past the point of adding shiny new features before Kilo is cut. I'd love to see a backlog spec for something like that to go into L | 22:28 |
| NobodyCam | ++ | 22:29 |
| rfchapman | JayF: ++ | 22:29 |
| *** hj-hp has quit IRC | 22:29 | |
| openstackgerrit | Nisha Agarwal proposed openstack/ironic: Guard inspect* fields for version < 1.6 https://review.openstack.org/165634 | 22:34 |
| rloo | Nisha: ? I already submitted a patch for that. https://review.openstack.org/#/c/165596/ | 22:35 |
| TheJulia | JayF: When you get a minute, please take a look at https://bugs.launchpad.net/ironic/+bug/1433812 | 22:36 |
| openstack | Launchpad bug 1433812 in Ironic "ironic-python-agent silently fails to write a configdrive if a previous drive is found" [Undecided,New] | 22:36 |
| Nisha | rloo, oh i tried to find it | 22:36 |
| Nisha | i didnt see so i submitted it | 22:36 |
| rloo | Nisha: sorry. I must have forgotten to ping you about it. | 22:36 |
| Nisha | anyway i will abandon mine | 22:36 |
| *** Marga_ has quit IRC | 22:36 | |
| JayF | TheJulia: nice, I'll bookmark that. I think there are multiple things at play here and maybe a race (i.e. it mounts the configdrive before it reads oem/cloud-config.yml) | 22:36 |
| *** absubram has quit IRC | 22:37 | |
| TheJulia | JayF: that is my feeling as well which is why I so verbose in what I found | 22:38 |
| mrda | Question on database migration - I need to change the length of the logical name field on a node to be 255 (from 63) chars. Can I just modify the existing alembic migration script? (i.e. we don't care about databases that have been created since logical_name landed) or should I keep the existing script (which added the field at 63 chars) and add a new migration script that ALTER TABLEs that up to | 22:40 |
| mrda | 255 (and hence support people who installed off HEAD)? | 22:40 |
| mrda | I probably think the second, but I don't want to add a new db migration as it slows things down on startup unless I need to | 22:42 |
| jlvillal | mrda: How long has the logical_name been landed? I also gut feeling would think the 2nd option. | 22:42 |
| *** absubram has joined #openstack-ironic | 22:45 | |
| mrda | jlvillal: yeah, it's been a while now - I'll just add it and people can comment in the review if they'd like different | 22:45 |
| *** Marga_ has joined #openstack-ironic | 22:47 | |
| *** andreykurilin_ has joined #openstack-ironic | 22:48 | |
| devananda | Nisha: i'm fixing up your patch now | 22:50 |
| Nisha | devananda, which one? | 22:50 |
| devananda | oh - never mind | 22:50 |
| devananda | rloo did the same things i was about to do | 22:51 |
| Nisha | devananda, :) | 22:51 |
| openstackgerrit | Josh Gachnang proposed openstack/ironic: Implement cleaning/zapping for the agent driver https://review.openstack.org/161453 | 22:52 |
| * BadCub goes wild-eyes and starts clicking links | 22:53 | |
| * devananda is bleary-eyed and headachy now too | 23:03 | |
| * JoshNang too | 23:05 | |
| NobodyCam | :( | 23:06 |
| *** subscope_ has quit IRC | 23:07 | |
| * NobodyCam offers asprin or rum based drinks! | 23:07 | |
| mrda | aspirin based drinks? :) | 23:12 |
| jlvillal | devananda: or anyone else: Is there something like a nova-liberty-proposed-sessions etherpad yet? I know it is over 2 months away but curious. | 23:12 |
| NobodyCam | mrda: lol | 23:13 |
| *** absubram has quit IRC | 23:14 | |
| devananda | jlvillal: not yet. I was going to start one in about 2 weeks | 23:15 |
| devananda | NobodyCam: or? *or*??! embrace "and" :-D | 23:16 |
| jlvillal | devananda: Thanks. | 23:16 |
| NobodyCam | devananda: We can do that too | 23:16 |
| NobodyCam | :) | 23:16 |
| devananda | https://etherpad.openstack.org/p/liberty-ironic-design-summit-ideas | 23:17 |
| devananda | have at it :) | 23:17 |
| * devananda dumps his notes on the pad, goes afk for a while | 23:18 | |
| jroll | NobodyCam: rum sounds good but I'll take coffee | 23:20 |
| jroll | :P | 23:20 |
| *** yuanying has joined #openstack-ironic | 23:23 | |
| *** mjturek3 has quit IRC | 23:25 | |
| *** andreykurilin_ has quit IRC | 23:25 | |
| *** Nisha has quit IRC | 23:25 | |
| *** Nisha has joined #openstack-ironic | 23:26 | |
| NobodyCam | :) | 23:26 |
| * BadCub ganders at new etherpad | 23:28 | |
| *** trown|call has joined #openstack-ironic | 23:29 | |
| *** trown has quit IRC | 23:29 | |
| * BadCub likes preemptive planning notes :-) | 23:29 | |
| * jroll writes some things | 23:32 | |
| NobodyCam | jroll: https://github.com/juliakreger/bifrost | 23:32 |
| jroll | right. | 23:33 |
| jroll | lol | 23:34 |
| NobodyCam | lol | 23:34 |
| jroll | I love etherpad conversations | 23:34 |
| NobodyCam | lol ++ | 23:34 |
| mrda | https://etherpad.openstack.org/p/liberty-ironic-design-summit-ideas | 23:36 |
| mrda | sorry paste mistake | 23:37 |
| NobodyCam | :) | 23:37 |
| jlvillal | NobodyCam: How about Server Trusted eXecution Technology (TXT). To help with server attestation. I wasn't sure if it was okay for me to just add it or not. | 23:38 |
| JayF | jlvillal: I think that's already got support in Ironic? | 23:38 |
| JayF | jlvillal: went in for K iirc | 23:38 |
| jlvillal | JayF: Oh, I guess I'm out of the loop, if so. | 23:38 |
| NobodyCam | jlvillal: add it these are things we will pick from.. we have limited slots | 23:39 |
| jlvillal | NobodyCam: Okay. | 23:39 |
| jlvillal | JayF: I'm not sure that it is in. I think it was proposed for Kilo | 23:39 |
| JayF | jlvillal: I can't find the spec. Maybe it didn't merge /me finding it | 23:39 |
| JayF | jlvillal: https://review.openstack.org/#/c/133902/ | 23:40 |
| *** hj-hp has joined #openstack-ironic | 23:40 | |
| JayF | jlvillal: so spec exists and made it through quite a few rounds of review, should be close for L I'd presume | 23:40 |
| jlvillal | JayF: Thanks | 23:40 |
| JayF | I say that without having read the spec recently though ... | 23:40 |
| openstackgerrit | Ruby Loo proposed openstack/ironic: Automate uefi boot iso creation for iscsi_ilo driver https://review.openstack.org/155900 | 23:50 |
| NobodyCam | TY rloo that looks much better :) | 23:52 |
| *** Nisha has quit IRC | 23:52 | |
| *** Nisha has joined #openstack-ironic | 23:52 | |
| jroll | oh man, I used two computers on that pad, now I have two colors | 23:55 |
| BadCub | lol | 23:55 |
| rloo | NobodyCam: yw. but i was wondering why i generated 'localhost' instead of 'ironic' for rabbit_host and qpid_hostname. any idea? | 23:56 |
| JoshNang | i ran into that earlier this week. localhost seems like the correct value | 23:57 |
| jroll | NobodyCam: what were the merge conflicts for https://review.openstack.org/#/c/157715 | 23:57 |
| *** jrist has quit IRC | 23:58 | |
| rloo | JoshNang: yeah, now that I think about it. localhost should be the right default. | 23:58 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
