*** wanyen has quit IRC | 00:14 | |
JayF | fyi; I just dropped a -2 on any remaining open ironic-specs reviews | 00:14 |
---|---|---|
openstackgerrit | Ellen Hui proposed a change to openstack/ironic: Make DHCP provider pluggable https://review.openstack.org/112351 | 00:27 |
*** jbjohnso_ has quit IRC | 00:27 | |
openstackgerrit | Josh Gachnang proposed a change to openstack/ironic: Improve IPA client library https://review.openstack.org/111118 | 00:28 |
*** jbjohnso_ has joined #openstack-ironic | 00:39 | |
*** jbjohnso_ has quit IRC | 00:54 | |
*** ellenh has quit IRC | 01:03 | |
*** jbjohnso_ has joined #openstack-ironic | 01:07 | |
*** nosnos has joined #openstack-ironic | 01:10 | |
*** dlaube has quit IRC | 01:18 | |
*** penick has quit IRC | 01:26 | |
*** eguz has joined #openstack-ironic | 01:53 | |
*** eghobo has quit IRC | 01:57 | |
*** Haomeng has joined #openstack-ironic | 01:59 | |
*** Krast has joined #openstack-ironic | 02:00 | |
openstackgerrit | Michael Davies proposed a change to openstack/ironic: Driver merge review comments from 111425-2-3 https://review.openstack.org/113750 | 02:14 |
mrda | Any Ironic eyeballs welcomed on ^^^. These are requested changes from Nova cores to the Ironic Driver, which need to be reviewed and approved, before we can move forward in Novaland. | 02:18 |
*** Krast has quit IRC | 02:23 | |
*** BrownBear has joined #openstack-ironic | 02:33 | |
*** Krast has joined #openstack-ironic | 02:42 | |
*** vinbs has joined #openstack-ironic | 03:03 | |
*** Poornima has joined #openstack-ironic | 03:05 | |
*** rainya has joined #openstack-ironic | 03:11 | |
*** ellenh has joined #openstack-ironic | 03:15 | |
*** rloo has quit IRC | 03:21 | |
*** shardy has quit IRC | 03:31 | |
*** nosnos has quit IRC | 03:35 | |
*** ellenh has quit IRC | 03:50 | |
*** dlaube has joined #openstack-ironic | 03:59 | |
*** eghobo has joined #openstack-ironic | 04:00 | |
*** dlaube has quit IRC | 04:01 | |
*** jroll|dupe has joined #openstack-ironic | 04:03 | |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Mechanism to cleanup all ImageCaches https://review.openstack.org/110560 | 04:05 |
*** jroll has quit IRC | 04:05 | |
*** jroll|dupe is now known as jroll | 04:05 | |
*** Poornima has quit IRC | 04:24 | |
*** rainya has quit IRC | 04:25 | |
*** nosnos has joined #openstack-ironic | 04:32 | |
*** eghobo has quit IRC | 04:37 | |
*** eghobo has joined #openstack-ironic | 04:40 | |
*** Poornima has joined #openstack-ironic | 04:48 | |
*** eghobo has quit IRC | 04:52 | |
*** jgrimm has quit IRC | 05:01 | |
*** bvivek has joined #openstack-ironic | 05:12 | |
*** Nisha has joined #openstack-ironic | 05:18 | |
*** k4n0 has joined #openstack-ironic | 05:30 | |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Take iSCSI deploy out of pxe driver https://review.openstack.org/111232 | 05:37 |
*** bmahalakshmi has joined #openstack-ironic | 05:45 | |
*** pcrews has quit IRC | 05:47 | |
openstackgerrit | Rakesh H S proposed a change to openstack/ironic: IPMI double bridging functionality https://review.openstack.org/95775 | 05:52 |
*** killer_prince is now known as lazy_prince | 05:55 | |
*** vinbs has quit IRC | 06:05 | |
*** dtantsur|afk is now known as dtantsur | 06:08 | |
dtantsur | Morning Ironic | 06:08 |
Zerpex | Morning :( | 06:09 |
dtantsur | Zerpex, morning, what happened? | 06:10 |
Zerpex | New day starts - trying to downgrade ironic :D | 06:11 |
Zerpex | dtantsur: do you know which version of ironic in git is compatible with Icehouse? | 06:19 |
dtantsur | Zerpex, hmmm... it's hard to figure out. | 06:24 |
dtantsur | lemme try | 06:25 |
dtantsur | Zerpex, chances are 4d4aeab4e57d623ca06bdfdceddbee2791851d03 is still compatible. I can't verify it though | 06:26 |
Zerpex | I'll take a look, thanks! | 06:26 |
*** sabah has joined #openstack-ironic | 06:26 | |
Zerpex | Found this commit https://github.com/openstack/ironic/commit/02417ebe8d506c1d6a0a7439e0352aed597470cf | 06:34 |
Zerpex | So I guess Ifa98e88c4c78ef904b79da57a47195344ff4b3fb will be what I need | 06:35 |
Zerpex | And is non existing.. FML | 06:35 |
dtantsur | Zerpex, it's not what you need I guess | 06:37 |
dtantsur | Zerpex, Icehouse branch is https://github.com/openstack/ironic/tree/stable/icehouse | 06:38 |
Zerpex | https://github.com/openstack/ironic/tree/stable/icehouse also seems like there's a stable branch for icehouse :o | 06:38 |
dtantsur | yes :) | 06:38 |
Zerpex | haha nice one | 06:38 |
dtantsur | Zerpex, commit ID that I posted previously is my attempt to guess the last _Juno_ commit that _might_ work with Icehouse Nova | 06:38 |
dtantsur | and btw things starting with I... like Ifa98e88c4c78ef904b79da57a47195344ff4b3fb are Gerrit change ID, not commit ID | 06:39 |
Zerpex | ohh | 06:40 |
*** jcoufal has joined #openstack-ironic | 06:40 | |
*** bvivek has quit IRC | 06:53 | |
*** lazy_prince is now known as killer_prince | 06:56 | |
openstackgerrit | A change was merged to openstack/python-ironicclient: Optimize get_by_instance_uuid https://review.openstack.org/113469 | 06:58 |
*** killer_prince is now known as lazy_prince | 06:59 | |
Zerpex | hmm, now I'm both running Ironic Icehouse + rest of openstack in Icehouse - but nova scheduler still gives http://snaps.lucasrolff.com/6a5035spaqwq8tn.png when I give it a restart :/ | 07:07 |
*** foexle has joined #openstack-ironic | 07:15 | |
*** Alexei_987 has joined #openstack-ironic | 07:27 | |
*** rwsu has quit IRC | 07:34 | |
*** ndipanov_gone is now known as ndipanov | 07:43 | |
*** rwsu has joined #openstack-ironic | 07:52 | |
openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Raise MissingParameterValue when validating glance info https://review.openstack.org/108456 | 07:53 |
*** derekh has joined #openstack-ironic | 08:04 | |
*** jistr has joined #openstack-ironic | 08:08 | |
*** bvivek has joined #openstack-ironic | 08:17 | |
*** ifarkas has joined #openstack-ironic | 08:20 | |
*** lucas-dinner is now known as lucasagomes | 08:24 | |
lucasagomes | jroll, hey... will take a look. When you pinged me it was really late here (~1am) :( so I didn't see | 08:24 |
* lucasagomes coffee | 08:25 | |
*** subah has joined #openstack-ironic | 08:31 | |
*** subah has quit IRC | 08:32 | |
*** subah has joined #openstack-ironic | 08:32 | |
*** sabah has quit IRC | 08:32 | |
*** bmahalakshmi has quit IRC | 08:33 | |
*** athomas has quit IRC | 08:33 | |
*** bmahalakshmi has joined #openstack-ironic | 08:34 | |
*** athomas has joined #openstack-ironic | 08:35 | |
*** athomas has quit IRC | 08:35 | |
*** athomas has joined #openstack-ironic | 08:36 | |
*** lazy_prince is now known as killer_prince | 08:38 | |
*** rameshg87 has joined #openstack-ironic | 08:44 | |
*** MattMan has quit IRC | 09:01 | |
*** MattMan has joined #openstack-ironic | 09:01 | |
rameshg87 | dtantsur, hi | 09:02 |
*** Jatin360 has joined #openstack-ironic | 09:04 | |
*** bvivek has quit IRC | 09:08 | |
*** rameshg87_ has joined #openstack-ironic | 09:15 | |
*** pelix has joined #openstack-ironic | 09:17 | |
*** rameshg87 has quit IRC | 09:18 | |
*** bmahalakshmi has quit IRC | 09:22 | |
*** Nisha has quit IRC | 09:37 | |
GheRivero | morning all | 09:53 |
mrda | Hi GheRivero | 09:55 |
lucasagomes | morning GheRivero mrda | 09:59 |
*** MattMan has quit IRC | 10:00 | |
*** SpamapS has quit IRC | 10:00 | |
*** zul has quit IRC | 10:00 | |
mrda | hey lucasagomes | 10:01 |
*** zul has joined #openstack-ironic | 10:01 | |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Take iSCSI deploy out of pxe driver https://review.openstack.org/111232 | 10:01 |
dtantsur | rameshg87_, hi! | 10:03 |
dtantsur | GheRivero, mrda, lucasagomes, morning/evening :) | 10:04 |
rameshg87_ | dtantsur, ^^^ | 10:04 |
rameshg87_ | dtantsur, hi | 10:04 |
dtantsur | rameshg87_, yeah I see | 10:04 |
rameshg87_ | dtantsur, i think i will have to do relative import in atleast one place | 10:04 |
rameshg87_ | dtantsur, and delay it until the opts are loaded | 10:05 |
mrda | hey dtantsur | 10:05 |
dtantsur | rameshg87_, I'm a bit confused, there should be not difference between absolute and relative import | 10:06 |
dtantsur | rameshg87_, also, relative imports won't work in (future) Oy3k | 10:06 |
dtantsur | Py3k actually | 10:06 |
dtantsur | rameshg87_, I still don't get why you can't just do 2 usual imports | 10:06 |
*** rameshg87__ has joined #openstack-ironic | 10:07 | |
dtantsur | rameshg87_, but import_opt is ok, the only question is what's the problem with absolute import | 10:07 |
*** MattMan has joined #openstack-ironic | 10:07 | |
*** SpamapS has joined #openstack-ironic | 10:07 | |
*** rameshg87_ has quit IRC | 10:09 | |
mrda | lucasagomes: jroll wants me to add https://review.openstack.org/108545 and https://review.openstack.org/113036 to the nova ironic patch series. But I'm only happy to do that if a) the pass tests (jroll: :-P) and b) a few cores agree with the changes. So if you don't mind taking a look, I think jroll would be happy. As for me, I'll come back online in 13 hours and see where those patches are at. | 10:09 |
*** rameshg87_ has joined #openstack-ironic | 10:09 | |
rameshg87_ | dtantsur, sorry keeps getting disconnected | 10:09 |
lucasagomes | mrda, I will take a look... if we think it's important to go in we also should check if the reviews on nova doesn't already have a +2 | 10:10 |
lucasagomes | I wouldn't like to override reviews with a +2 :( | 10:10 |
lucasagomes | I've reviewed the first patch, there's a py3 incompatible problem there | 10:10 |
mrda | lucasagomes: We won't be doing that :) But I'll add them on the end of the patch series as "improvements" so as to be able to keep the +2's | 10:11 |
mrda | I don't want to make things harder, or discourage, nova reviewers | 10:11 |
lucasagomes | mrda, ah, I see... isn't it something that jroll can do himself? (add to the end of the patch series) | 10:12 |
lucasagomes | and that will be a review in nova | 10:12 |
dtantsur | lucasagomes, +1 to having independent review in nova | 10:12 |
dtantsur | that's how we'll work in the future anyway :) | 10:12 |
lucasagomes | yeah | 10:12 |
*** rameshg87__ has quit IRC | 10:12 | |
mrda | well, deva has sworn everyone away from the proposed patches, so I'll happily do it. | 10:12 |
mrda | That way I can keep on top of them | 10:12 |
mrda | :) | 10:13 |
mrda | as in "this is the official patch series we want landed in Juno, there's no extra cruft, only what needs to be included at this time" | 10:13 |
dtantsur | rameshg87_, another suggestion (not sure): can we move definitions of the options to iscsi_deploy.py without changing them (thus staying backward compatible) | 10:14 |
lucasagomes | I see... well I think the cache is important, but if it's only going to be added to the end of the patch series or going to override a +2 in nova... I think we should wait the series to get merged and them we can propose it in nova directly | 10:14 |
mrda | lucasagomes: then it won't land in J | 10:15 |
rameshg87_ | dtantsur, ah okay, still retaining the same group name ? | 10:15 |
dtantsur | rameshg87_, yeah | 10:15 |
lucasagomes | mrda, right... but it wouldn't anyway right? by adding it to the end I mean it's very unlikely that it's going to be merged | 10:15 |
mrda | basically, outside of what we deem as important, won't land because 3 weeks | 10:15 |
dtantsur | rameshg87_, this will keep you from creating circular imports (which work, but I'd better do w/o them) | 10:15 |
rameshg87_ | dtantsur, devananda raised concern on requiring users to change config files | 10:15 |
mrda | I can only hope :) | 10:15 |
rameshg87_ | dtantsur, if we do this, that is address | 10:15 |
dtantsur | rameshg87_, that won't change config, if you keep the group and name | 10:16 |
dtantsur | (it will change comment in the example config, which is not relevant to backward compatibility) | 10:16 |
*** jcoufal has quit IRC | 10:19 | |
lucasagomes | mrda, btw, patch series #5 there's a -1 from dan about https://review.openstack.org/#/c/113047/ | 10:19 |
lucasagomes | which is now merged in ironic | 10:19 |
mrda | thanks lucasagomes | 10:19 |
*** rameshg87_ has quit IRC | 10:20 | |
*** bvivek has joined #openstack-ironic | 10:21 | |
*** dtantsur is now known as dtantsur|lunch | 10:23 | |
*** MattMan has quit IRC | 10:23 | |
*** SpamapS has quit IRC | 10:23 | |
*** MattMan has joined #openstack-ironic | 10:30 | |
*** SpamapS has joined #openstack-ironic | 10:30 | |
*** bmahalakshmi has joined #openstack-ironic | 10:33 | |
*** rameshg87 has joined #openstack-ironic | 10:36 | |
rameshg87 | dtantsur|lunch, we cannot move all definitions to iscsi_deploy.py | 10:42 |
*** subah has quit IRC | 10:43 | |
rameshg87 | dtantsur|lunch, for example image_cache_ttl and image_cache_size for now is shared between iscsi_deploy.py and pxe.py | 10:43 |
rameshg87 | dtantsur|lunch, so i guess we will have to import for atleast two of these at one place | 10:43 |
rameshg87 | dtantsur|lunch, or move all definitions to iscsi_deploy and do import from pxe. let me try that. | 10:44 |
rameshg87 | dtantsur|lunch, and circular import with absolution "from X import Y" doesn't seem to work | 10:44 |
rameshg87 | dtantsur|lunch, this fails: http://paste.openstack.org/show/94314/ | 10:45 |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Take iSCSI deploy out of pxe driver https://review.openstack.org/111232 | 10:59 |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE Installation Guide documentation https://review.openstack.org/106809 | 11:00 |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add iPXE Installation Guide documentation https://review.openstack.org/106809 | 11:02 |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Add support for creating vfat disk images https://review.openstack.org/105413 | 11:05 |
*** rameshg87 has quit IRC | 11:06 | |
*** rameshg87 has joined #openstack-ironic | 11:06 | |
*** jcoufal has joined #openstack-ironic | 11:09 | |
*** derekh has quit IRC | 11:09 | |
*** rakesh_hs has joined #openstack-ironic | 11:09 | |
*** rameshg87 has quit IRC | 11:17 | |
*** bvivek2 has joined #openstack-ironic | 11:21 | |
*** bvivek has quit IRC | 11:22 | |
*** lucasagomes is now known as lucas-hungry | 11:22 | |
*** rakesh_hs has quit IRC | 11:28 | |
*** rameshg87 has joined #openstack-ironic | 11:31 | |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: Add support for interacting with swift https://review.openstack.org/105795 | 11:34 |
*** Poornima has quit IRC | 11:41 | |
*** nosnos has quit IRC | 11:54 | |
*** BrownBear has quit IRC | 12:04 | |
*** Jatin360 has quit IRC | 12:11 | |
*** boris-42 has quit IRC | 12:21 | |
openstackgerrit | Ramakrishnan G proposed a change to openstack/ironic: IloVirtualMediaIscsi deploy driver https://review.openstack.org/113865 | 12:22 |
*** lucas-hungry is now known as lucasagomes | 12:22 | |
*** bmahalakshmi has quit IRC | 12:31 | |
*** boris-42 has joined #openstack-ironic | 12:32 | |
*** rameshg87 has quit IRC | 12:34 | |
*** rameshg87 has joined #openstack-ironic | 12:35 | |
*** rameshg87 has quit IRC | 12:48 | |
*** matty_dubs|gone is now known as matty_dubs | 12:56 | |
*** aswadr has joined #openstack-ironic | 13:02 | |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add list() to Chassis, Node, Port objects https://review.openstack.org/113877 | 13:14 |
jroll | morning ironic | 13:14 |
jroll | :) | 13:14 |
jroll | lucasagomes: no worries on pinging you late, I just ping people whenever, and let them deal with it when they're around :) | 13:15 |
lucasagomes | jroll, :D sure | 13:15 |
jroll | just like when you ponged it was 1:30 here :P | 13:15 |
lucasagomes | hah | 13:15 |
lucasagomes | yeah TZ's sucks :/ | 13:15 |
jroll | mrda, I feel like that cache needs to land in juno... ironic is fairly unusable without it at any sort of scale | 13:16 |
jroll | yep | 13:16 |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add list() to Chassis, Node, Port objects https://review.openstack.org/113877 | 13:18 |
*** dtantsur|lunch is now known as dtantsur | 13:26 | |
dtantsur | jroll, morning | 13:27 |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add list() to Chassis, Node, Port objects https://review.openstack.org/113877 | 13:28 |
*** rloo has joined #openstack-ironic | 13:29 | |
*** bvivek2 has quit IRC | 13:32 | |
jroll | morning dtantsur :) | 13:32 |
*** pcrews has joined #openstack-ironic | 13:42 | |
dtantsur | hey, anyone here with HP box can call `sudo ipmitool lan print`? | 13:52 |
dtantsur | we're collecting some statistics :) | 13:52 |
*** jasondotstar has joined #openstack-ironic | 13:56 | |
*** igordcard has joined #openstack-ironic | 13:58 | |
*** linggao has joined #openstack-ironic | 14:05 | |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add list() to Chassis, Node, Port objects https://review.openstack.org/113877 | 14:05 |
lucasagomes | >.<' | 14:05 |
openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Query full node details and cache https://review.openstack.org/108545 | 14:11 |
openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Use cache in node_is_available() https://review.openstack.org/113036 | 14:11 |
jroll | lucasagomes: ^ addressed your comments | 14:11 |
jroll | not sure yet why tempest was failing, I'll take a look a bit later | 14:12 |
lucasagomes | jroll, ta much! will take a look I gotta a call pretty soon | 14:12 |
lucasagomes | take a look soon* | 14:12 |
jroll | no worries, I'm going to head to the office and whatnot so I'll be afk for a bit | 14:12 |
lucasagomes | ack | 14:12 |
*** Krast has quit IRC | 14:13 | |
*** Krast has joined #openstack-ironic | 14:14 | |
*** openstackgerrit has quit IRC | 14:16 | |
NobodyCam | happy hump day Ironic | 14:17 |
*** openstackgerrit has joined #openstack-ironic | 14:18 | |
openstackgerrit | A change was merged to openstack/ironic: Raise MissingParameterValue instead of Invalid https://review.openstack.org/108455 | 14:19 |
*** pradipta_away is now known as pradipta | 14:20 | |
lucasagomes | NobodyCam, morning | 14:24 |
NobodyCam | morning lucasagomes :) | 14:24 |
lucasagomes | NobodyCam, I was thinking about u after I saw a very weird bug ticket (not in ironic) | 14:24 |
lucasagomes | I know you like odd bugs hah | 14:24 |
lucasagomes | lemme find the link | 14:24 |
Shrews | omg... i totally misread that | 14:24 |
lucasagomes | NobodyCam, https://bugs.launchpad.net/ubuntu/+source/cupsys/+bug/255161/comments/28 | 14:24 |
lucasagomes | check that out ^ | 14:24 |
Shrews | "NobodyCam, I was thinking about u after I saw a very weird bug" | 14:24 |
lucasagomes | Shrews, u too :P | 14:24 |
NobodyCam | lol morning Shrews :) | 14:25 |
lucasagomes | hah, I mean weird == funny :) | 14:25 |
lucasagomes | "open office will never print on Tuesdays" | 14:25 |
Shrews | i read it as bug == 6 legged insect | 14:25 |
NobodyCam | hahhahaha | 14:25 |
NobodyCam | Shrews: 6 legs much better then 8 | 14:26 |
NobodyCam | :-p | 14:26 |
Shrews | well, yeah. one less pair of shoes to buy | 14:26 |
lucasagomes | heh | 14:27 |
*** Krast has quit IRC | 14:28 | |
Shrews | wow, that is a neat bug | 14:28 |
*** Krast has joined #openstack-ironic | 14:28 | |
NobodyCam | that is a neat bug | 14:28 |
NobodyCam | lol | 14:28 |
NobodyCam | raining quite a lot this morning up in north WA :-p | 14:30 |
*** Krast has quit IRC | 14:31 | |
*** Krast has joined #openstack-ironic | 14:31 | |
NobodyCam | lol : https://www.linkedin.com/today/post/article/20140809132357-1270637-an-honest-message-from-your-it-guy | 14:36 |
openstackgerrit | A change was merged to openstack/ironic: Mechanism to cleanup all ImageCaches https://review.openstack.org/110560 | 14:38 |
*** foexle has quit IRC | 14:46 | |
*** matty_dubs has quit IRC | 14:50 | |
*** jbjohnso__ has joined #openstack-ironic | 14:57 | |
*** ndipanov has quit IRC | 14:59 | |
NobodyCam | rloo: do you have a second to take another look at https://review.openstack.org/#/c/108456 | 15:00 |
rloo | looking... | 15:00 |
rloo | NobodyCam: so I don't understand why Ghe wants to make this one change. | 15:00 |
rloo | NobodyCam: there are other places that use InvalidParameterValue too. Why just change this one? | 15:01 |
*** jgrimm has joined #openstack-ironic | 15:01 | |
NobodyCam | looks 108455 looks like it got the bulk of them | 15:01 |
*** jbjohnso_ has quit IRC | 15:01 | |
rloo | NobodyCam: so MissingParameterValue is subclassed from InvalidParameterValue. Which means that if you use InvalidParameterValue, it is OK. | 15:02 |
NobodyCam | would you be better if there was a bug to track any reamaining ones? | 15:02 |
NobodyCam | I took it as providing a clear error to the operator | 15:03 |
rloo | NobodyCam: well, I'd like some consistency. I wonder whether subclassing makes things muddier. | 15:03 |
*** ndipanov has joined #openstack-ironic | 15:04 | |
rloo | NobodyCam: I did a quick test by unsubclassing MissingParameterValue, and the unit tests failed in various places. So why just modify one of those places? | 15:04 |
rloo | NobodyCam: I'm not saying that the change is wrong, I would like to understand the reasoning for that one change vs changing them all. | 15:05 |
NobodyCam | I can understand that. I will see if I can get GheRivero to post a comment as to the reasoning for it | 15:06 |
rloo | NobodyCam: notice that others have asked too, why he didn't make the change in 108455, and I don't see any reply. | 15:07 |
NobodyCam | rloo: ya, I saw that. but when I looked 108455 had landed | 15:08 |
rloo | NobodyCam: yeah, I approved 108455 today ;) | 15:09 |
NobodyCam | :) | 15:10 |
dtantsur | folks, let me retry please: <dtantsur> hey, anyone here with HP box can call `sudo ipmitool lan print`? | 15:13 |
lucasagomes | NobodyCam, ^ | 15:17 |
lucasagomes | dtantsur, maybe also some people on TripleO with access to the tripleo cloud | 15:17 |
lucasagomes | dtantsur, I think they can access the boxes there | 15:18 |
NobodyCam | dtantsur: I have a hp laptop with ubuntu I run that command on ... wil that work for you? | 15:18 |
lucasagomes | dtantsur, check with derekh on #tripleo, he's on top of CI stuff | 15:18 |
dtantsur | NobodyCam, only if it has BMC running :) likely no | 15:18 |
lucasagomes | NobodyCam, I think he mean a HP iLO server | 15:18 |
NobodyCam | ya | 15:18 |
NobodyCam | :( not me sorry :( | 15:19 |
*** romcheg1 has joined #openstack-ironic | 15:21 | |
openstackgerrit | Imre Farkas proposed a change to openstack/ironic: Add DracDriver and its DracPower module https://review.openstack.org/104850 | 15:36 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Use valid exceptions in start_shellinabox_console https://review.openstack.org/113939 | 15:43 |
NobodyCam | is there a rule about copyright dates if / when you split functions from one file to a new one.. ie do we keep the orginal copyright date, use current date, or add current to the new file so it has both? | 15:50 |
lucasagomes | NobodyCam, hmm good q... I'm not sure, but I think that it would be nice to copy the copyright string as-is over to the new file | 15:51 |
*** ndipanov has quit IRC | 15:52 | |
NobodyCam | lucasagomes: thats what they did.. I was thinking we should add a new header for the current date too? example https://review.openstack.org/#/c/111232/9/ironic/drivers/modules/iscsi_deploy.py | 15:53 |
lucasagomes | NobodyCam, hmm... well, like updating the date? if they updated the code yeah it may make sense | 15:54 |
lucasagomes | tho I wouldn't care much about it, this copyright lines are all optional | 15:55 |
lucasagomes | and all the code belongs to the openstack foundation anyway (you signed the CLA) | 15:55 |
lucasagomes | so | 15:55 |
NobodyCam | ya just looked odd seeing new files with old dates | 15:55 |
NobodyCam | :-p | 15:55 |
NobodyCam | wouldn't -1 it for that | 15:55 |
NobodyCam | just looked odd to me | 15:55 |
lucasagomes | yeah, but anyway... i don't know if there's an methodology for that | 15:56 |
lucasagomes | I would say that if u updated the code when moved it would makes sense to update the date | 15:56 |
lucasagomes | if you just moved I would leave the date as is | 15:56 |
lucasagomes | because git tracks content and not files anyway | 15:56 |
lucasagomes | so the content is the same as when it was written | 15:56 |
NobodyCam | ahh good point | 15:56 |
Shrews | https://wiki.openstack.org/wiki/LegalIssuesFAQ#Copyright_Headers | 15:59 |
lucasagomes | jroll, this is a theory... when I use the cache patch my machines won't PXE boot. And I think that's because the nova driver will update the node with the VIF information, but when it gets from the cache that info may not be there... so it fails to update neutron and the thing won't boot | 15:59 |
lucasagomes | and that's my theory why the tempest is not passing | 15:59 |
NobodyCam | oh great find Shrews TY | 15:59 |
*** eghobo has joined #openstack-ironic | 15:59 | |
*** dlaube has joined #openstack-ironic | 16:00 | |
*** k4n0 has quit IRC | 16:01 | |
Shrews | ugh. gate is amazingly unstable lately | 16:02 |
NobodyCam | lately = last couple of months? | 16:02 |
NobodyCam | :( | 16:02 |
Shrews | well, yesterday and today especially | 16:03 |
lucasagomes | Shrews, if you have time https://review.openstack.org/#/c/113276/ this is a tempest patch for the management interface stuff | 16:03 |
lucasagomes | the last one | 16:03 |
Shrews | lucasagomes: looking... | 16:03 |
lucasagomes | if you get a time to review it, it's quite small | 16:03 |
lucasagomes | after that we can mark the blueprint as completed | 16:03 |
rloo | yuriyz: wrt your comment for 113939 -- good catch. should that be in a separate patch though? | 16:04 |
Shrews | lucasagomes: bah. i thought i reviewed that yesterday? sorry, guess i got distracted and forgot to vote | 16:04 |
lucasagomes | Shrews, heh, thanks! | 16:04 |
*** ndipanov has joined #openstack-ironic | 16:04 | |
rloo | yuriyz: never mind, I'm going to make the change anyway and see if anyone objects. thx! | 16:04 |
yuriyz | hi rloo ok :) | 16:05 |
jroll | lucasagomes: you might be right, I like that theory | 16:12 |
jroll | I hate it because it means my code is wrong, but | 16:12 |
jroll | ya know. | 16:12 |
lucasagomes | heh yeah | 16:13 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Use valid exception in start_shellinabox_console https://review.openstack.org/113939 | 16:13 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Use valid exception in start_shellinabox_console https://review.openstack.org/113939 | 16:14 |
*** ellenh has joined #openstack-ironic | 16:16 | |
NobodyCam | morning jroll :) | 16:20 |
*** jcoufal has quit IRC | 16:20 | |
jroll | morning NobodyCam :) | 16:24 |
openstackgerrit | Ellen Hui proposed a change to openstack/ironic: Make DHCP provider pluggable https://review.openstack.org/112351 | 16:28 |
devananda | mornin, all | 16:33 |
NobodyCam | good morning devananda | 16:33 |
rloo | morning devananda | 16:34 |
dtantsur | devananda, morning | 16:34 |
jroll | morning deva | 16:35 |
*** romcheg1 has quit IRC | 16:40 | |
openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Remove objectify decorator from dbapi.update_* methods https://review.openstack.org/113968 | 16:41 |
lucasagomes | devananda, morning | 16:41 |
jroll | lucasagomes: so on your theory... where does the nova driver update neutron? all of the (un)plug_vifs calls do a node.get | 16:47 |
jroll | afaict | 16:47 |
lucasagomes | jroll, hmm wait... the ironic nova driver doesn't update neutron it updates the nodes in ironic with the VIF infos... | 16:49 |
lucasagomes | jroll, lemme try to dig more into it here, I got the env | 16:49 |
devananda | nova compute makes those calls | 16:49 |
jroll | yeah | 16:49 |
devananda | it gets the MAC addr from ironic | 16:49 |
lucasagomes | oh | 16:49 |
devananda | get_macs_for_node | 16:49 |
devananda | trace back to what calls that | 16:50 |
jroll | lucasagomes: regardless, there's a node.get at the beginning of spawn() | 16:50 |
JayF | Would most folks in here consider it a bug that Ironic uses power status to validate power status was changed but doesn't use get boot device similarly? | 16:50 |
jroll | lucasagomes: I think there is a nova thing that does hit neutron, however | 16:50 |
* jroll digs | 16:50 | |
lucasagomes | devananda, right, just to give you a context... the cache patch that jroll is working on is making my machines to fail to PXE boot, so I'm trying to look around the VIF stuff on the driver. And the theory was that the driver will update the ironic node with the VIF info, but when it gets that node again from the cache that info may not be there | 16:51 |
lucasagomes | but I'm not sure... yeah lemme dig more into it | 16:51 |
jroll | lucasagomes: I'm also curious, because patch set 2 passed tempest and not a lot changed, just the refresh=True, really | 16:52 |
lucasagomes | JayF, power status to validate power status? | 16:52 |
jroll | also deallocate_networks_on_reschedule landed | 16:52 |
jroll | lucasagomes: https://review.openstack.org/#/c/108545/2..6/ironic/nova/virt/ironic/driver.py | 16:52 |
lucasagomes | JayF, setting the power/provision state is async in Ironic, where setting the boot device is not | 16:53 |
NobodyCam | devananda: got's a second to take a look at https://review.openstack.org/#/c/113750 (has three +2's) but is to nova driver so no one is +a'ing :-p | 16:53 |
devananda | lucasagomes: ahh. right ... | 16:53 |
lucasagomes | JayF, if we want to make it async, yeah I would think that we need a target_boot_device etc... | 16:53 |
JayF | lucasagomes: Let me ask the question in a less generic way then: Is it reasonable for me to want Ironic to validate the BMC did what it said it did w/r/t setting a boot device. With power status we confirm this with a loop, by running it after we set power state, etc | 16:54 |
JayF | lucasagomes: I'm seeing cases where I have a strong suspicion that something failed -- most likely in the bmc -- when setting a boot device... but with power these cases are caught because power status doesn't match what it's supposed to be | 16:54 |
lucasagomes | JayF, right... well if it fails in the BMC I would expect the ipmi command to fail... but let's assume it failed but the ipmi command succeed in run, you still can do a get_boot_device() to confirm that the request device is the one set on the BMC | 16:55 |
lucasagomes | now I don't know if that should be the default behavior :/ | 16:56 |
lucasagomes | unless we _can't_ really trust any of the ipmi commands to say it ran successfully when it didn't | 16:56 |
JayF | lucasagomes: We have a specific failure case on our BMCs where 'power off' commands can be ignored. i.e. you tell a node to power off, and it just stays on. | 16:57 |
devananda | brb | 16:57 |
JayF | That failure case is fixed on our nodes now, but I'm just seeing behavior that makes me suspect the same could be happening with boot device | 16:57 |
JayF | but ironic doesn't check any of that right now... I think it'd certainly be interesting to make boot device state work like power state | 16:58 |
JayF | where we know what it should be, sync it up, etc | 16:58 |
lucasagomes | JayF, yeah, I think that the idea of having a fully async api in K is to make everything that talks to the BMC like the power state we have today | 16:58 |
lucasagomes | all async with target states etc | 16:59 |
lucasagomes | so yeah, that's urgh... BMCs seems to be super problematic and not trustable at all :( (I've almost no hands on experience with it, so it's good to have feedback from guys like u) | 17:01 |
*** chuckC has quit IRC | 17:01 | |
*** bmahalakshmi has joined #openstack-ironic | 17:03 | |
JayF | for most people bmc is a service of last resort | 17:03 |
JayF | i.e. "man I'm glad I have this BMC so I don't have to drive to the datacenter" | 17:03 |
JayF | not people writing software suites to use them to manipulate hardware :) | 17:04 |
jroll | I want opencompute hardware with an open source ilo controller | 17:04 |
lucasagomes | :D | 17:04 |
lucasagomes | ic, so I think we should work hard on K to get the fully async API and convert everything that touches the BMC to a same/similar model of the power state stuff | 17:05 |
*** krtaylor has quit IRC | 17:06 | |
jroll | lucasagomes: when your servers fail to pxe, are you sure they're turning on at all? | 17:07 |
jroll | lucasagomes: in CI logs I don't see a deploy request :| | 17:07 |
jroll | I think it might be a resource tracker issue | 17:07 |
lucasagomes | jroll, yup, it's power on with a "no more network devices" error message from pxe | 17:08 |
jroll | oh hm | 17:08 |
* jroll spins up devstack | 17:08 | |
lucasagomes | 4 baremetalbrbm_0 running | 17:08 |
jroll | cool | 17:08 |
lucasagomes | virsh output ^ | 17:08 |
lucasagomes | virsh list* | 17:09 |
*** MattMan has left #openstack-ironic | 17:09 | |
* lucasagomes fetches the last version of the cache patch and re-run devstack | 17:12 | |
*** Alexei_987 has quit IRC | 17:17 | |
*** romcheg1 has joined #openstack-ironic | 17:22 | |
lucasagomes | jroll, ew wait... it just PXE booted here with the last version... >.< | 17:22 |
jroll | yay, transient bugs. | 17:22 |
lucasagomes | :/ | 17:22 |
jroll | I have a new patch, we'll let tempest run on that | 17:22 |
dlaube | JayF: I'm curious about what hardware vendor/BMC you guys are using | 17:22 |
dlaube | RE "BMCs where 'power off' commands can be ignored" | 17:22 |
JayF | dlaube: It's a bug in our specific revision of the ME we're using. | 17:23 |
dlaube | ahh ok | 17:23 |
openstackgerrit | Jim Rollenhagen proposed a change to openstack/ironic: Query full node details and cache https://review.openstack.org/108545 | 17:23 |
dlaube | we are about to buy some hardware and want to use ironic for our provisioning | 17:23 |
dlaube | I'm wondering if one vendor is better supported than others | 17:24 |
JayF | dlaube: I would not personally speak on that, because that's not my job at Rackspace so much | 17:24 |
dlaube | or if it mainly boils down to wonky IPMI/BMC stuff between the different types of hardware out there | 17:24 |
JayF | dlaube: but I will philosophically endorse the use of OpenCompute design servers | 17:24 |
NobodyCam | mrda: are you around? | 17:25 |
JayF | Pretty much. I think it's a matter of pick your quirks, not sure picking one without quirks is an option :P | 17:25 |
jroll | dlaube: there's a hot new ilo driver coming down the pipe, I like that from a software standpoint but I've never used it :P | 17:25 |
dlaube | gotcha | 17:25 |
dlaube | heheh | 17:25 |
jroll | dlaube: most out-of-band stuff sucks, btw | 17:25 |
jroll | as far as I can tell | 17:25 |
jroll | brb | 17:25 |
dlaube | I've been following this -> https://review.openstack.org/#/c/95775/ | 17:26 |
*** ndipanov has quit IRC | 17:26 | |
dlaube | in hopes of using more dense hardware and controlling via BMC | 17:26 |
* lucasagomes gotta review that patch | 17:29 | |
lucasagomes | from the last review I did I think it was quite OK, so I think it may land soon in ironic | 17:30 |
lucasagomes | and that's priority as well since we have a blueprint for that work | 17:30 |
*** eguz has joined #openstack-ironic | 17:31 | |
lucasagomes | jroll, ohhhhh you may be right about the resource tracking man | 17:31 |
dlaube | lucasagomes: that's good to hear | 17:32 |
*** igordcard has quit IRC | 17:32 | |
lucasagomes | jroll, I usually kill n-cpu to speed up the resource tracking stuff | 17:32 |
*** romcheg1 has quit IRC | 17:32 | |
lucasagomes | jroll, http://paste.openstack.org/show/94524/ | 17:33 |
lucasagomes | jroll, now I can see that tho my node has a valid power state (that's what the driver looks at before returning back the resources status) the tracker still reporting 0 CPUs, DISK, etc... | 17:33 |
lucasagomes | that may cause tempest to timeout | 17:34 |
NobodyCam | brb | 17:35 |
*** eghobo has quit IRC | 17:35 | |
lucasagomes | brb too, it's late here... jroll I will look more into it later/tomorrow | 17:35 |
lucasagomes | have a good night everybody | 17:35 |
devananda | dlaube: reviews and comments to the effect of "hey, i tested this on $my-hardware and it works" are great. so are the opposite (hey, it broke!) | 17:36 |
devananda | g'night lucas! | 17:36 |
lucasagomes | night! | 17:36 |
*** lucasagomes is now known as lucas-dinner | 17:36 | |
*** pradipta is now known as pradipta_away | 17:37 | |
dlaube | devananda: I'll be sure to chime in here with some feedback once we get rolling with ironic | 17:38 |
devananda | dlaube: if you dont mind, i'm curious - who do yo uwork for, and what size deployment are you expecting soon(tm)? | 17:39 |
dlaube | I sure wish there was a supported/tested hardware doc out there | 17:39 |
dlaube | working for a small host and we are looking to get our feet wet with ironic. We've previously used homebrewed provisioning systems to interact with PXE booting, imaging servers, etc | 17:40 |
dlaube | we'll be deploying to a lab environment soonish | 17:40 |
devananda | coolness | 17:41 |
devananda | for hardware - anything with IPMI support should work | 17:41 |
devananda | i'd stay away from supermicro - their IPMI security is fairly poor - if that's sometjhing you care about | 17:42 |
devananda | most major vendors should implement the common things that ironic needs | 17:43 |
dlaube | sweet | 17:43 |
devananda | HP is a major contributor, so you can probably expect good hardware support (disclaimer: i work for them, too) | 17:43 |
dlaube | yeah, we are not glued to one vendor or another but we were looking at supermicro | 17:43 |
dlaube | hehe, thank you for the honesty ;) | 17:43 |
devananda | watch out for their low-end stuff. anything that puts the IPMI controller on a shared data port is *bad* for security | 17:44 |
*** pelix has quit IRC | 17:44 | |
jroll | lucas-dinner: hmm, I'll check it out. thanks for the info, have a good night :) | 17:44 |
dlaube | devananda: are you referring to those boxes that only have two NICs for *everything*? As opposed to two NICs for public/private traffic and a third just dedicated for IPMI LAN ? | 17:47 |
devananda | dlaube: yes | 17:47 |
*** dlaube has quit IRC | 17:48 | |
devananda | dlaube: "those boxes" have a hidden MAC on the first data port, which internally maps to the IPMI controller | 17:48 |
devananda | tldr; you need a dedicated IPMI LAN | 17:48 |
devananda | if the hardware doesn't support that, don't buy it | 17:48 |
jroll | 10:48:03 <-- | dlaube (~Adium@2601:9:6a80:3e3:d0ba:4c13:9668:ae8d) has quit (Read error: Connection reset by peer) | 17:49 |
openstackgerrit | A change was merged to openstack/ironic: Add list() to Chassis, Node, Port objects https://review.openstack.org/113877 | 17:51 |
*** chuckC has joined #openstack-ironic | 17:54 | |
*** dlaube has joined #openstack-ironic | 17:56 | |
*** penick has joined #openstack-ironic | 18:00 | |
*** penick has quit IRC | 18:00 | |
*** penick has joined #openstack-ironic | 18:02 | |
*** jistr has quit IRC | 18:08 | |
NobodyCam | brb ... quick walkies in the rain :-p | 18:14 |
* dlaube forgets what rain walkies feel like after moving to the bay area | 18:15 | |
jroll | heh | 18:15 |
jroll | dlaube: you missed this: | 18:16 |
jroll | 10:48:19 devananda | dlaube: "those boxes" have a hidden MAC on the first data port, which internally maps to the IPMI controller | 18:16 |
jroll | 10:48:32 devananda | tldr; you need a dedicated IPMI LAN | 18:16 |
jroll | 10:48:39 devananda | if the hardware doesn't support that, don't buy it | 18:16 |
jroll | sorry deva for the pings :/ | 18:16 |
dlaube | ahhh | 18:16 |
dlaube | thank you! | 18:16 |
JayF | dlaube: yeah, dedicated BMC port is one feature I'd look for, and good on deva for thinking of it when I didn't :) | 18:16 |
dlaube | good stuff | 18:17 |
jroll | who in here can get devstack cores to review things, because apparently I can't | 18:18 |
*** lucasagomes_ has joined #openstack-ironic | 18:35 | |
Shrews | jroll: have you pinged people directly? | 18:36 |
Shrews | i've found general pleas for help will get ignored. you need to go right to the source | 18:36 |
*** Hefeweiz1n has joined #openstack-ironic | 18:36 | |
Shrews | source(s) | 18:37 |
*** lucas-dinner has quit IRC | 18:41 | |
*** openstackgerrit has quit IRC | 18:41 | |
*** Hefeweizen has quit IRC | 18:41 | |
*** ellenh has quit IRC | 18:41 | |
*** penick has quit IRC | 18:45 | |
*** penick has joined #openstack-ironic | 18:46 | |
*** jistr has joined #openstack-ironic | 18:47 | |
*** openstackgerrit has joined #openstack-ironic | 18:50 | |
*** penick has quit IRC | 18:52 | |
*** aswadr has quit IRC | 18:52 | |
*** eguz has quit IRC | 18:56 | |
*** eghobo has joined #openstack-ironic | 19:02 | |
*** penick has joined #openstack-ironic | 19:03 | |
*** bmahalakshmi has quit IRC | 19:08 | |
*** ellenh has joined #openstack-ironic | 19:12 | |
jroll | Shrews: idk who to ping :( | 19:12 |
NobodyCam | jroll: https://review.openstack.org/#/admin/groups/50,members | 19:13 |
jroll | how do you get to those groups, I can never find them :( | 19:13 |
jroll | ah I see | 19:13 |
* jroll bookmarks | 19:13 | |
NobodyCam | https://review.openstack.org/#/admin/groups/ | 19:14 |
NobodyCam | :) | 19:14 |
jroll | now the trick is mapping to irc nicks :P | 19:14 |
NobodyCam | thats another story :-p | 19:15 |
jroll | wheeee, got two | 19:15 |
jroll | sdague seems to be offline | 19:15 |
jroll | which is weir | 19:16 |
jroll | d | 19:16 |
Shrews | jroll: he should be your last resort | 19:16 |
jroll | oh? why? | 19:17 |
jroll | Shrews: he's just super busy or? | 19:17 |
Shrews | James "sdague" Brown? Hardest working man in Openstack | 19:17 |
Shrews | :) | 19:17 |
Shrews | yeah, just has his hands in everything. | 19:18 |
Shrews | busy busy | 19:18 |
jroll | lol | 19:18 |
ellenh | jroll: could you take another look at https://review.openstack.org/#/c/112351/ when you have time? Finally passing tests. :D | 19:21 |
jroll | ellenh: sure, gimme a few | 19:23 |
NobodyCam | jroll: looks like a new version of 111223 was pushed up with out landing 105590 :-p | 19:26 |
jroll | wtf | 19:33 |
*** dtantsur is now known as dtantsur|afk | 19:33 | |
NobodyCam | :( | 19:36 |
NobodyCam | night dtantsur|afk | 19:36 |
jroll | and now it has a +2 | 19:37 |
jroll | so mrda won't update it again | 19:37 |
jroll | lovely | 19:37 |
*** dlaube has quit IRC | 19:37 | |
NobodyCam | :( | 19:37 |
jroll | mrda: if you change https://review.openstack.org/#/c/111223/ please please please add in https://review.openstack.org/#/c/105590 | 19:37 |
jroll | huh, this doesn't look right :/ | 19:38 |
*** eghobo has quit IRC | 19:42 | |
*** eghobo has joined #openstack-ironic | 19:42 | |
NobodyCam | maybe more please's? | 19:43 |
jroll | PLEASE | 19:43 |
jroll | :P | 19:43 |
NobodyCam | lol | 19:44 |
*** eghobo has quit IRC | 19:45 | |
NobodyCam | GheRivero: if you get a free minute maind rebasing https://review.openstack.org/#/c/103685 | 19:47 |
NobodyCam | s/maind/mind/ | 19:47 |
*** krtaylor has joined #openstack-ironic | 19:49 | |
Shrews | ellenh: any particular reason you defined a __init__ in the dhcp base class that does nothing? i'm not seeing a need for that. | 19:50 |
Shrews | should only need the abstract methods | 19:50 |
ellenh | Shrews: ah, good point. It used to have a concrete method, i just forgot to remove __init__ | 19:52 |
Shrews | ellenh: :) i left a comment | 19:53 |
ellenh | Shrews: thanks! :) | 19:53 |
NobodyCam | oh this is a new one: http://logs.openstack.org/31/103331/3/check-tripleo/check-tripleo-ironic-undercloud-precise-nonha/ec9b74e/console.html.gz#_2014-08-08_23_58_57_534 | 19:57 |
openstackgerrit | Ellen Hui proposed a change to openstack/ironic: Make DHCP provider pluggable https://review.openstack.org/112351 | 20:00 |
*** jistr has quit IRC | 20:16 | |
GheRivero | NobodyCam: on my way to rebase https://review.openstack.org/#/c/103685 | 20:30 |
NobodyCam | GheRivero: Awesome Thank you :) | 20:32 |
*** ellenh1 has joined #openstack-ironic | 20:44 | |
*** ellenh has quit IRC | 20:46 | |
*** linggao has quit IRC | 20:46 | |
NobodyCam | brb | 20:49 |
*** stevebaker has quit IRC | 20:52 | |
*** stevebaker has joined #openstack-ironic | 20:52 | |
*** datajerk1 has quit IRC | 20:55 | |
*** ellenh1 has quit IRC | 20:57 | |
*** ellenh1 has joined #openstack-ironic | 21:01 | |
*** eghobo has joined #openstack-ironic | 21:02 | |
mrda | Morning Ironic! | 21:08 |
jroll | morning mrda | 21:10 |
* jroll looks around for something sharp :) | 21:10 | |
mrda | orly? | 21:10 |
* mrda reads scrollback | 21:10 | |
JayF | jroll: point those scissors away from your monitor | 21:10 |
NobodyCam | good morning mrda | 21:11 |
NobodyCam | lol | 21:11 |
NobodyCam | good afternoon JayF :) | 21:11 |
mrda | hey NobodyCam | 21:11 |
mrda | Looks like my late night push wasn't quite right | 21:11 |
NobodyCam | mrda: I had a question but i've forgotten it now..lol :-p | 21:12 |
mrda | umm, so NobodyCam, can you please +2 +A https://review.openstack.org/#/c/105590 | 21:12 |
mrda | I'll push a new 111223 this morning and include jroll's patch | 21:12 |
*** krtaylor has quit IRC | 21:13 | |
NobodyCam | mrda: yep ..TY :) | 21:13 |
mrda | (if you're happy doing so) | 21:13 |
NobodyCam | i think I +2'd it already | 21:13 |
mrda | nah, lucas did | 21:13 |
*** krtaylor has joined #openstack-ironic | 21:14 | |
*** datajerk has joined #openstack-ironic | 21:14 | |
jroll | mrda: I know there's already a +2 on whatever that's rolling into, so if you don't want to apply it that's fine | 21:14 |
NobodyCam | oh I thought I did | 21:15 |
NobodyCam | let me run a recheck odd nova error in check-tripleo-ironic-undercloud-precise-nonha | 21:15 |
mrda | jroll: I just want the ironic side to be up to date with what is going into Nova | 21:16 |
jroll | mrda: I see | 21:16 |
*** jasondotstar has quit IRC | 21:26 | |
*** igordcard has joined #openstack-ironic | 21:32 | |
*** krtaylor has quit IRC | 21:46 | |
NobodyCam | quick run to grab a pack of smokes... brb | 21:46 |
jroll | lifeless: I think there are plenty of reasons to support "not using neutron" for dhcp :/ | 21:52 |
jroll | lifeless: there are deployers that choose not to use neutron | 21:53 |
jroll | lifeless: there are deployers with existing dhcp infrastructure | 21:53 |
jroll | etc etc | 21:53 |
lifeless | jroll: thats like deployers that choose not to use keystone | 21:53 |
jroll | lifeless: there are also methods of configuring a deploy ramdisk that are not dhcp-based, as well, this could support that | 21:53 |
lifeless | it doesn't really make sense because they can just have a no-op keystone | 21:53 |
jroll | I tend to disagree | 21:54 |
lifeless | I'm fine with non-dhcp ramdisks | 21:54 |
lifeless | thats orthogonal | 21:54 |
lifeless | existing dhcpservers - sure, thats where you replace the dnsmasq backend in the dhcp aagent | 21:54 |
lifeless | have the agent talk to isc-dhcp or whatever | 21:54 |
lifeless | but it doesn't make technical sense to have multiple copies of the abstraction point | 21:54 |
lifeless | hell, they can deploy the neutron-server and dhcp-agent on the same machine as a conductor, write their abstraction point there, and away they go | 21:55 |
lifeless | no bigger implications | 21:55 |
jroll | so for people that want a single static configuration for ramdisk dhcp, they should write a no-op neutron dhcp driver? | 21:55 |
lifeless | let me check I understand the scenario | 21:56 |
lifeless | different l2 network for deploying vs deployed | 21:56 |
lifeless | and in the deploying scenario they're happy for wildcard dhcp ? | 21:56 |
lifeless | s/happy/desirous of|whatever | 21:57 |
jroll | not wildcard necessarily, but not controlled by ironic | 21:57 |
lifeless | so sure, in fact they don't need to do anything special, the runtime neutron being on a different l2 network will have no impact | 21:57 |
jroll | for the deployment network, I don't want ironic to control dhcp | 21:58 |
JayF | lifeless: in our case specifically; our DHCP configs are built using teh same automation from our 'cmdb' source that we use to add nodes to ironic | 21:58 |
jroll | and I don't want to run neutron on that network | 21:58 |
JayF | lifeless: so we aren't running wildcard dhcp fwiw | 21:58 |
lifeless | jroll: right, thats what I just said | 21:58 |
*** penick has quit IRC | 21:58 | |
lifeless | jroll: Ironic doesn't need to know that its not doing dhcp on the deployment network as we don't have any assumption about source IP address during deployment | 21:59 |
jroll | so you're saying to make a bunch of api calls to neutron that don't actually affect anything | 21:59 |
lifeless | jroll: they would be setting it up for deployed status | 21:59 |
lifeless | jroll: which is actually important ;) | 22:00 |
jroll | I'm not sure I'm grokking what you're saying here | 22:00 |
jroll | ironic has nothing to do with dhcp for a provisioned server | 22:00 |
lifeless | jroll: sure does | 22:00 |
JayF | jroll: it does for pxe nodes | 22:00 |
lifeless | jroll: perhaps not in onmetal | 22:01 |
jroll | oh | 22:01 |
jroll | right | 22:01 |
JayF | lifeless: not in ipa driver, you mean | 22:01 |
jroll | so there are drivers that don't pxe boot | 22:01 |
jroll | the instance image | 22:01 |
jroll | in tree | 22:01 |
lifeless | JayF: IPA will need to support pxe boot too eventually | 22:01 |
jroll | agree | 22:01 |
lifeless | JayF: so I don't see it being a driver issue | 22:01 |
lifeless | its a user choice | 22:01 |
lifeless | if they supply a full disk image | 22:01 |
JayF | A user should be able to choose to deploy ironic without Neutron, I fully agree :) | 22:01 |
lifeless | vs a AMI/AKI/ARI tuple | 22:01 |
jroll | heh | 22:01 |
jroll | right, that's the main issue here | 22:01 |
jroll | is that ironic should be able to be run/deployed without neutron | 22:02 |
jroll | without nova | 22:02 |
jroll | without keystone | 22:02 |
lifeless | but why | 22:02 |
lifeless | I get without nova | 22:02 |
lifeless | if you don't want quotas etc | 22:02 |
jroll | because monolithic software sucks | 22:02 |
lifeless | totally fine | 22:02 |
jroll | because neutron sucks | 22:02 |
JayF | A secondary issue is whether or not we should spend time talking about the suitability of a feature after the spec has not only been approved, but spec approval deadline has passed :( | 22:02 |
* JayF thought we had finished with this part of the discussion and moved on to implementing things | 22:02 | |
lifeless | jroll: I'm not promoting monolithism | 22:04 |
lifeless | jroll: I'm promoting using the API as the abstraction since thats what its *there for* | 22:04 |
lifeless | jroll: rather than adding a lot of code to implement the same abstraction in Ironic | 22:04 |
jroll | requiring me to run a specific service, at all, is promoting monolithism IMO | 22:04 |
JayF | requiring me to write plugins to Neutron is promoting cruel and unusual punishment IMO ;) | 22:05 |
jroll | ironic should be coupled to as little software as possible | 22:05 |
lifeless | jroll: APIs provide loose coupling - particularly good ones | 22:05 |
lifeless | jroll: arguing that neutron sucks -> so lets help make neutron suck less | 22:05 |
jroll | except that it becomes another service that I need to deploy, operate, and HA | 22:06 |
lifeless | jroll: I know plenty of folk that would love an isc-dhcp backend for the dhcp agent | 22:06 |
jroll | sure | 22:06 |
jroll | I would like that as well | 22:06 |
jroll | but I wouldn't run it | 22:06 |
jroll | and I don't know dhcp well enough to implement it | 22:06 |
jroll | and that makes for a really bad developer of a feature | 22:06 |
lifeless | why wouldn't you run it? I thought Rackspace had deployed Neutron now? | 22:06 |
jroll | (I might run it, idk) | 22:07 |
jroll | we have | 22:07 |
jroll | it's also an operational PITA for us :) | 22:07 |
lifeless | so look, I'm not going to block this - if I was I would have -2'd it. | 22:07 |
lifeless | But it doesn't make any technical sense to me. | 22:07 |
jroll | sure | 22:08 |
jroll | I happen to disagree and wanted to have the conversation | 22:08 |
lifeless | The broad use cases ('I want to deploy with a different infrastructure and deployer network'), 'I want a different DHCP server under the hood' etc do. | 22:08 |
jroll | I do agree that it's a fine line, I think I'm just slightly on the other side | 22:08 |
lifeless | My objection is that whenever we add abstraction points that duplicate those elsewhere in OpenStack, we split effort | 22:08 |
lifeless | and we end up with bloated hard to maintain software | 22:09 |
JayF | I would make the exact argument against what you're advocating for :) | 22:09 |
jroll | right, I see your point | 22:09 |
lifeless | like | 22:10 |
JayF | That having external services/apis adds a significant maintenance cost | 22:10 |
lifeless | we wouldn't add a thing to Ironic to teach it to create VMs | 22:10 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Use valid exception in start_shellinabox_console https://review.openstack.org/113939 | 22:10 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Improve exception handling in console code https://review.openstack.org/114046 | 22:10 |
lifeless | now, I think its likely an *entirely* fair argument to say that DHCP management via API would make sense as a standalone service much smaller than Neutron | 22:11 |
lifeless | and I'd support ripping it out of Neutron into such a mini or even micro service. | 22:11 |
lifeless | because Neutron is fat (as is Nova). | 22:12 |
jroll | agree | 22:12 |
jroll | can I write it in twisted? | 22:12 |
lifeless | thats not my call :/ | 22:12 |
jroll | I'm half joking | 22:13 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Use valid exception in start_shellinabox_console https://review.openstack.org/113939 | 22:13 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Improve exception handling in console code https://review.openstack.org/114046 | 22:14 |
NobodyCam | rloo: just a commit message update for 113939 | 22:15 |
rloo | NobodyCam: yes, I accidentally reverted the commit msg change in 113939, so added it back again. All for a missing 'w' :-) | 22:16 |
NobodyCam | lol I saw that | 22:16 |
NobodyCam | hehehehe | 22:16 |
jroll | uh oh, when did twisted get removed from global-requirements.txt? :( | 22:17 |
*** rloo is now known as rloo_afk | 22:19 | |
jroll | lifeless: https://github.com/psychomario/PyPXE :) | 22:19 |
jroll | specifically https://github.com/psychomario/PyPXE/blob/master/pypxe/dhcp.py | 22:19 |
lifeless | jroll: cool | 22:20 |
*** rwsu has quit IRC | 22:21 | |
*** penick has joined #openstack-ironic | 22:23 | |
SpamapS | Whats the word on retrying a 'deploy_failed' node? | 22:25 |
jroll | SpamapS: do you really want to know? :| | 22:26 |
jroll | update nodes set provision_state = NULL, target_provision_state = NULL where uuid = '$uuid'; | 22:27 |
jroll | afaik | 22:27 |
JayF | s/NULL/decommissioning/ if you're running the agent and live in the future | 22:27 |
jroll | -.- | 22:28 |
mrda | NobodyCam: looks like 105590 passed. Feel free to +2 +A :) | 22:29 |
jroll | woot | 22:30 |
mrda | jroll: once that process *starts* I'll push a new 111223 | 22:31 |
jroll | mrda: no worries, just let me know if I can help with anything | 22:31 |
jroll | :) | 22:31 |
NobodyCam | mrda: jroll: 105590 on its way down hte pipe :) | 22:32 |
jroll | wheeeeee | 22:32 |
NobodyCam | :) | 22:33 |
jroll | one less patch on my dashboard | 22:33 |
NobodyCam | only if it passes gate tests :-p | 22:33 |
* JayF rolls 2d6 | 22:33 | |
NobodyCam | lol | 22:33 |
JayF | Not craps! Looks like we pass the gate this time | 22:33 |
NobodyCam | JayF: maybe we should use 2d10 | 22:34 |
mrda | thanks NobodyCam - will go get you a free drink in Paris :) | 22:34 |
NobodyCam | lol :) | 22:34 |
*** rwsu has joined #openstack-ironic | 22:36 | |
*** penick has quit IRC | 22:37 | |
*** penick has joined #openstack-ironic | 22:37 | |
SpamapS | jroll: so.. Kilo? ;) | 22:38 |
jroll | SpamapS: I hope so :( | 22:38 |
jroll | there's so many kilo specs I want to submit already | 22:39 |
*** penick has quit IRC | 22:39 | |
SpamapS | For now, if we fail to PXE boot a compute node.. ah so sorry user, your instance go bye bye | 22:40 |
dguerri`afk | hello there! Is this: http://s27.postimg.org/3kn2i9d1u/Screen_Shot_2014_08_13_at_11_37_42_pm.jpg a known bug? | 22:40 |
*** dguerri`afk is now known as dguerri | 22:40 | |
dguerri | it seem my baremetal node is using more cpu than it has | 22:41 |
dguerri | :) | 22:41 |
dguerri | I expected the scheduler selected the second node | 22:42 |
jroll | SpamapS: yes | 22:44 |
*** jgrimm has quit IRC | 22:44 | |
jroll | dguerri: interesting, also what dashboard is that? | 22:44 |
dguerri | horizon, from devstack (Hypervisor page) | 22:45 |
jroll | ah yeah | 22:45 |
dguerri | actually just a portion of that... | 22:45 |
dguerri | :) | 22:45 |
jroll | yeah, I haven't seen that, but I doubt most folks look at their hypervisor stats for ironic nodes :) | 22:46 |
dguerri | hehe | 22:46 |
dguerri | I was just playing with the scheduler and I noticed that | 22:46 |
JayF | comstud: ^ didn't we see this? a node that got double-provisioned? I remember avail_ram showing up as -32GB on one of our compute nodes once | 22:46 |
jroll | ah yeah | 22:46 |
comstud | the -32GB is a side effect of... | 22:46 |
comstud | putting a node in maintenance while it has an instance | 22:46 |
dguerri | I will fill a bug report, but I would like to be sure it not.. me | 22:47 |
comstud | that's all | 22:47 |
comstud | what happens is... | 22:47 |
jroll | dguerri: it might be you, but yeah go ahead :) | 22:47 |
comstud | the driver reports the node has 0 memory | 22:47 |
comstud | and nova subtracts the instance usage from that | 22:47 |
JayF | his is the opposite though | 22:47 |
JayF | nova says it's provisioned 2 cpus on a node that only has 1 | 22:47 |
JayF | so probably unrelated? | 22:47 |
comstud | i have never seen that.. the resource tracker should prevent 2 instances from going to same node | 22:48 |
jroll | oh, the node's properties could have changed | 22:48 |
jroll | if it had 2 at provision time | 22:48 |
comstud | sorry, gotta step away for a few | 22:48 |
jroll | and now it has one | 22:48 |
jroll | dguerri: ^^ feasible? | 22:48 |
comstud | yeah, that could happen | 22:48 |
comstud | maybe | 22:48 |
SpamapS | jroll: at the very least, we need to stop turning power off on a box when it errors | 22:48 |
jroll | that is a known bug | 22:48 |
SpamapS | jroll: I've lost the console because of that.. so I don't even know why it failed :-P | 22:48 |
dguerri | jroll nope | 22:48 |
jroll | SpamapS: why, you actually want to debug things? :) | 22:48 |
mrda | jroll: you should be happy now :) | 22:49 |
dguerri | jroll I have created a node with 2 cpus just to try out the scheduling | 22:49 |
jroll | SpamapS: I had to hot-patch that code when we initially started spinning up our environment | 22:49 |
jroll | mrda: thank you! | 22:49 |
jroll | mrda: I have more patches to get through, but there's a resource tracker bug in them | 22:49 |
jroll | comstud: when you're back maybe you want to help me with that ^ :) | 22:50 |
mrda | jroll: 3 more commits to go to get things back to normal :-/ | 22:50 |
dguerri | the problem is that the scheduler selected the wrong node.. | 22:50 |
jroll | dguerri: oh, are you using the exactcpufilter or whatever? | 22:50 |
dguerri | could that be something wrong with the flavor I am using? | 22:50 |
jroll | or that | 22:50 |
dguerri | exactcpufilter ! | 22:50 |
dguerri | seems interesting | 22:51 |
dguerri | what is that? :) | 22:51 |
jroll | yes | 22:51 |
jroll | sec :) | 22:51 |
dguerri | it makes sense | 22:51 |
jroll | I mean, it shouldn't choose one with not enough cores | 22:51 |
jroll | but yeah | 22:52 |
dguerri | checking... | 22:52 |
dguerri | jroll do you have a link handy? | 22:53 |
comstud | Yeah, I was just coming back to say.. | 22:53 |
jroll | dguerri: | 22:53 |
jroll | scheduler_available_filters=ironic.nova.scheduler.default_filters.default_filters | 22:53 |
jroll | scheduler_default_filters=RetryFilter,ComputeFilter,ExactRamFilter,ExactCoreFilter | 22:53 |
comstud | If you don't ahve an ExactCoreFilter, that would also do it | 22:53 |
jroll | is what's in our environment | 22:53 |
NobodyCam | nice error: http://logs.openstack.org/36/96136/14/check/check-tempest-dsvm-virtual-ironic/39f6a36/console.html.gz#_2014-07-23_22_46_13_000 | 22:54 |
jroll | comstud: but it looked for 2 CPUs and got one afaik | 22:54 |
jroll | which is... odd | 22:54 |
comstud | jroll: that nova.scheduler.default_filters.default_filters may only be in our branch | 22:54 |
jroll | comstud: heh | 22:54 |
jroll | I was wondering, I didn't think I saw that before | 22:54 |
comstud | as a hack to avoid a puppet issue we have | 22:54 |
jroll | yeah | 22:54 |
* jroll looks for real ironic docs | 22:54 | |
comstud | [15:53:44] <jroll> scheduler_default_filters=RetryFilter,ComputeFilter,ExactRamFilter,ExactCoreFilter | 22:55 |
comstud | that's correct | 22:55 |
jroll | ah ok | 22:55 |
comstud | if you pull the ironic stuff into the nova tree | 22:55 |
comstud | and you don't have to change scheduler_available_filters | 22:55 |
comstud | but if you run with the filters out of tree, you have to modify the available filters | 22:55 |
jroll | yeah, that's the problem :P | 22:55 |
comstud | (which is what we do right now) | 22:55 |
comstud | one may want to add the ExactDiskFilter too or whatever | 22:56 |
comstud | but it's not necessary for us.. at least atm | 22:56 |
jroll | right | 22:56 |
comstud | the Ram Filter actually covers us completely | 22:56 |
comstud | heh | 22:56 |
*** chuckC has quit IRC | 22:58 | |
*** chuckC has joined #openstack-ironic | 22:58 | |
*** penick has joined #openstack-ironic | 22:59 | |
openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Avoid calling _parse_deploy_info twice https://review.openstack.org/108442 | 23:01 |
*** chuckC has quit IRC | 23:01 | |
*** lucasagomes_ has quit IRC | 23:03 | |
openstackgerrit | A change was merged to openstack/ironic: Allow Ironic URL from config file https://review.openstack.org/105590 | 23:07 |
NobodyCam | jroll: ^^^^^ | 23:07 |
JayF | jroll: ^ woo | 23:07 |
mrda | glad that happened since I prposed it in the driver :P | 23:09 |
mrda | oh, btw ironic cores, can I get a +A on https://review.openstack.org/#/c/113750/ ? | 23:10 |
*** dguerri is now known as dguerri`afk | 23:11 | |
mrda | I'd like to propose this into nova today | 23:11 |
mrda | NobodyCam Shrews jroll | 23:12 |
mrda | ^^^ | 23:12 |
openstackgerrit | Ghe Rivero proposed a change to openstack/ironic: Fix tear_down a node with missing info https://review.openstack.org/103685 | 23:13 |
NobodyCam | mrda: +a'd | 23:14 |
NobodyCam | GheRivero: Awesome TY | 23:14 |
mrda | thanks NobodyCam - much appreciated! | 23:14 |
openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic: Updated from global requirements https://review.openstack.org/114057 | 23:19 |
openstackgerrit | OpenStack Proposal Bot proposed a change to openstack/ironic-python-agent: Updated from global requirements https://review.openstack.org/114058 | 23:19 |
*** BrownBear has joined #openstack-ironic | 23:21 | |
jroll | motherfffffffffff | 23:32 |
jroll | I found the issue with 108545 | 23:32 |
jroll | the cache is populated before power state is grabbed | 23:32 |
jroll | I also wonder if the resource tracker loop isn't running, or if it doesn't refresh the cache | 23:33 |
jroll | I think the latter | 23:33 |
NobodyCam | anyone seen this one: No distributions matching the version for oslo.config>=1.4.0.0a3 | 23:35 |
jroll | comstud: ^ I decided to use the refresh parameter to get_available_nodes, resource tracker loop doesn't send True for that | 23:36 |
*** chuckC has joined #openstack-ironic | 23:37 | |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Use valid exception in start_shellinabox_console https://review.openstack.org/113939 | 23:37 |
jroll | NobodyCam: yeah, upgrade your pip I think | 23:37 |
openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Improve exception handling in console code https://review.openstack.org/114046 | 23:37 |
jroll | NobodyCam: and/or pass --use-wheel? iirc | 23:37 |
*** ellenh1 has quit IRC | 23:38 | |
jroll | hmmmm, not sure how I want to work around this :| | 23:39 |
*** radsy has joined #openstack-ironic | 23:39 | |
*** igordcard has quit IRC | 23:40 | |
NobodyCam | jroll: upgraded from pip 1.2.1 to 1.5.6 and still have same error :( | 23:47 |
jroll | NobodyCam: try with --use-wheel | 23:47 |
jroll | or something like that | 23:47 |
openstackgerrit | A change was merged to openstack/ironic: Raise MissingParameterValue when validating glance info https://review.openstack.org/108456 | 23:49 |
openstackgerrit | A change was merged to openstack/ironic: Driver merge review comments from 111425-2-3 https://review.openstack.org/113750 | 23:49 |
NobodyCam | jroll: add that to pip or tox command? | 23:52 |
jroll | NobodyCam: pip | 23:52 |
jroll | but not entirely sure | 23:52 |
NobodyCam | oh neat... http://paste.openstack.org/show/94707/ | 23:54 |
NobodyCam | looks like the py2.py3 tag is throwing me for a curve | 23:55 |
jroll | NobodyCam: hrm, idk | 23:59 |
jroll | probably want to ask infra | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!