| *** michchap has joined #openstack-ironic | 00:05 | |
| *** anniec has quit IRC | 00:06 | |
| *** anniec has joined #openstack-ironic | 00:10 | |
| *** michchap_ has joined #openstack-ironic | 00:30 | |
| *** matsuhashi has joined #openstack-ironic | 00:30 | |
| *** michchap has quit IRC | 00:33 | |
| *** michchap_ has quit IRC | 00:44 | |
| *** rloo has quit IRC | 00:45 | |
| *** rloo has joined #openstack-ironic | 00:46 | |
| *** michchap has joined #openstack-ironic | 00:47 | |
| *** nosnos has joined #openstack-ironic | 01:03 | |
| *** sjing has joined #openstack-ironic | 01:40 | |
| *** sjing has quit IRC | 02:07 | |
| *** sjing has joined #openstack-ironic | 02:09 | |
| *** anniec has quit IRC | 02:26 | |
| *** michchap has quit IRC | 02:48 | |
| *** lnxnut has joined #openstack-ironic | 02:58 | |
| *** anteaya has quit IRC | 03:00 | |
| *** rloo has quit IRC | 03:12 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: 123456789012345678903234567890423456789123456789 Changes power_state and adds last_error field https://review.openstack.org/54466 | 03:16 |
|---|---|---|
| *** prekarat has joined #openstack-ironic | 03:23 | |
| *** prekarat has quit IRC | 03:24 | |
| *** lnxnut has quit IRC | 03:42 | |
| *** lnxnut has joined #openstack-ironic | 03:43 | |
| openstackgerrit | Ruby Loo proposed a change to openstack/ironic: Changes power_state and adds last_error field https://review.openstack.org/54466 | 03:46 |
| *** lnxnut has quit IRC | 03:47 | |
| *** harlowja has quit IRC | 04:07 | |
| *** jianingy_afk has quit IRC | 04:18 | |
| *** sjing has quit IRC | 04:47 | |
| *** sjing has joined #openstack-ironic | 04:48 | |
| *** prekarat has joined #openstack-ironic | 05:51 | |
| *** blamar has quit IRC | 06:14 | |
| *** martyntaylor has joined #openstack-ironic | 06:34 | |
| *** sjing has quit IRC | 06:42 | |
| *** sjing has joined #openstack-ironic | 06:44 | |
| *** Haomeng|2 has quit IRC | 07:20 | |
| *** Haomeng has joined #openstack-ironic | 07:39 | |
| *** prekarat has quit IRC | 07:41 | |
| *** prekarat has joined #openstack-ironic | 07:54 | |
| *** romcheg has joined #openstack-ironic | 08:00 | |
| GheRivero | morning ironic! | 08:08 |
| Haomeng | Morning GheRivero:) | 08:19 |
| *** prekarat has quit IRC | 08:33 | |
| *** sjing has quit IRC | 08:49 | |
| *** matsuhashi has quit IRC | 09:03 | |
| *** matsuhashi has joined #openstack-ironic | 09:03 | |
| *** derekh has joined #openstack-ironic | 09:05 | |
| *** jistr has joined #openstack-ironic | 09:06 | |
| *** matsuhas_ has joined #openstack-ironic | 09:08 | |
| *** matsuhashi has quit IRC | 09:08 | |
| lifeless | devanand1: | 09:09 |
| lifeless | devanand1: https://bugs.launchpad.net/bugs/1237802 being 'opinion' surprises me | 09:09 |
| *** romcheg has quit IRC | 09:13 | |
| Haomeng | lifeless: how are you | 09:28 |
| Haomeng | I think devanand1 is away status | 09:28 |
| *** prekarat has joined #openstack-ironic | 09:31 | |
| *** lucasagomes has joined #openstack-ironic | 09:32 | |
| *** romcheg has joined #openstack-ironic | 09:35 | |
| romcheg | lifeless: are you around? | 09:36 |
| romcheg | Morning Haomeng and everyone else | 09:39 |
| *** prekarat has quit IRC | 09:50 | |
| *** prekarat has joined #openstack-ironic | 09:59 | |
| romcheg | Hi lucasagomes, are you around? | 10:21 |
| lucasagomes | romcheg, hey yes | 10:22 |
| lucasagomes | how r u? | 10:22 |
| romcheg | I'm good, thanks. Yourself? | 10:22 |
| lucasagomes | not too bad :) sleepy | 10:22 |
| romcheg | I'm just looking to the client and trying to make it compatible with py3 | 10:24 |
| lucasagomes | :D, I think there was a guy submitting a couple of patches around it while ago | 10:24 |
| romcheg | from ironicclient.openstack.common.py3kcompat import urlutils | 10:25 |
| lucasagomes | https://review.openstack.org/#/q/status:merged+project:openstack/python-ironicclient,n,z | 10:25 |
| lucasagomes | Kui Shi his name | 10:25 |
| romcheg | I noticed that this file duplicates six.moves.urlparse | 10:25 |
| romcheg | six.moves.urllib even | 10:25 |
| romcheg | http://pythonhosted.org/six/#module-six.moves.urllib.parse | 10:26 |
| romcheg | Do you agree that it's necessary to remove it? | 10:27 |
| Haomeng | morning, Romcheg and lucasagomes and everyelse:) | 10:32 |
| max_lobur | morning Everyone :) | 10:35 |
| lucasagomes | romcheg, to remove the py3kcompat.urlutils layer? That's what other projects r using as well no? | 10:36 |
| lucasagomes | Haomeng, max_lobur morning | 10:36 |
| romcheg | lucasagomes: Other projects use it, but the same functionality is already implemented in six | 10:37 |
| lucasagomes | romcheg, ah got what u mean (sorry I'm sleepy and slow), so if the layer is not needed anymore because it's already in six I would drop it yea, but we have to make sure we use the right version of six | 10:43 |
| romcheg | ok then | 10:43 |
| romcheg | I also see that some of the py3 failures are related to keystoneclient | 10:44 |
| romcheg | I will also try to figure that out | 10:44 |
| lucasagomes | yea, that's the reason why we have p3 non-voting now | 10:45 |
| Haomeng | yes, i noticednow,but I have concern, how to make sure all codes are Compatible both p2 and p3 at the same time | 10:48 |
| Haomeng | so we will keep one eye about 3rd libs | 10:49 |
| *** prekarat has quit IRC | 10:51 | |
| *** matsuhas_ has quit IRC | 11:01 | |
| *** matsuhashi has joined #openstack-ironic | 11:02 | |
| *** prekarat has joined #openstack-ironic | 11:07 | |
| *** prekarat has quit IRC | 11:08 | |
| *** matsuhashi has quit IRC | 11:11 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/python-ironicclient: Add driver-list https://review.openstack.org/53683 | 11:13 |
| openstackgerrit | Yuriy Zveryanskyy proposed a change to openstack/ironic: Add wrapper for keystone service catalog https://review.openstack.org/52601 | 11:22 |
| *** matsuhashi has joined #openstack-ironic | 12:02 | |
| *** jbjohnso has quit IRC | 12:06 | |
| *** lucasagomes is now known as lucas-lunch | 12:12 | |
| *** nosnos has quit IRC | 12:27 | |
| *** jdob has joined #openstack-ironic | 12:27 | |
| Haomeng | lucasagomes:ping | 12:54 |
| *** linggao has joined #openstack-ironic | 12:58 | |
| *** anteaya has joined #openstack-ironic | 13:00 | |
| *** fpavlik has joined #openstack-ironic | 13:00 | |
| *** fpavlik has quit IRC | 13:03 | |
| Haomeng | hi Ironic, good morning/evening:) | 13:04 |
| Haomeng | who can help to confirm if our unittest support the real database, such as database exceptions? I understand we use fake db in ut | 13:04 |
| Haomeng | I am writing ut code, found it does not support real database exceptions | 13:05 |
| *** rloo has joined #openstack-ironic | 13:07 | |
| *** matsuhashi has quit IRC | 13:07 | |
| *** matsuhashi has joined #openstack-ironic | 13:08 | |
| *** matsuhas_ has joined #openstack-ironic | 13:09 | |
| *** matsuhashi has quit IRC | 13:10 | |
| openstackgerrit | Haomeng,Wang proposed a change to openstack/ironic: Change Port create API to EAFP https://review.openstack.org/54537 | 13:13 |
| *** lucas-lunch is now known as lucasagomes | 13:15 | |
| *** rloo has quit IRC | 13:20 | |
| *** rloo has joined #openstack-ironic | 13:21 | |
| *** jbjohnso has joined #openstack-ironic | 13:21 | |
| *** rloo has quit IRC | 13:23 | |
| *** rloo has joined #openstack-ironic | 13:24 | |
| lucasagomes | hi Haomeng | 13:39 |
| lucasagomes | you have to mock it | 13:39 |
| lucasagomes | Haomeng, lemme test here locally and I show you how to do it | 13:40 |
| Haomeng | thank you lucasagomes | 13:40 |
| Haomeng | mock a database exception? | 13:41 |
| jbjohnso | devanand1, I can probably soon reserve 0x34 command and then nest bridging so that something like moonshot can be supported, but I have no way to test the beast | 13:42 |
| Haomeng | lucasagomes, if you are busy on other things, you can just give me a reference, I will try | 13:43 |
| Haomeng | this ismy patch -https://review.openstack.org/#/c/54537/ | 13:44 |
| lucasagomes | Haomeng, it's alright I'm just commenting on ur patch how to do it | 13:44 |
| lucasagomes | so you can go and update it urself | 13:44 |
| Haomeng | ok | 13:45 |
| *** rloo has quit IRC | 13:45 | |
| Haomeng | thank you:) | 13:45 |
| Haomeng | ok | 13:45 |
| jbjohnso | devanand1, though I wonder if they actually have a reason to specify local_ipmb_address or not, hard to imagine a scenario where 0x20 isn't the value | 13:45 |
| Haomeng | I will do bymyself, thankyou Lucas:) | 13:45 |
| *** rloo has joined #openstack-ironic | 13:45 | |
| lucasagomes | Haomeng, commented | 13:46 |
| Haomeng | got it, thank you:) | 13:46 |
| lucasagomes | Haomeng, np, so there you mocked the second save() forcing it to raise the desired exception | 13:47 |
| Haomeng | yes, cool:) | 13:47 |
| Haomeng | will try your solution onmytomorrow morning:) | 13:48 |
| Haomeng | what isyour time now? | 13:48 |
| lucasagomes | 13:48 | 13:48 |
| Haomeng | I am tired today, will go tosleep first, good afternoon:) | 13:48 |
| Haomeng | :) | 13:48 |
| lucasagomes | Haomeng, ack! have a good night! | 13:48 |
| Haomeng | my timeis 9:48pm:) | 13:48 |
| Haomeng | :) | 13:49 |
| lucasagomes | ^^ yea it's pretty late there | 13:49 |
| lucasagomes | enjoy ur night I see ya tomorrow | 13:49 |
| Haomeng | today I am very tired, for other days,will work around11pm | 13:49 |
| Haomeng | ok | 13:49 |
| Haomeng | :) | 13:49 |
| * devanand1 waves | 14:11 | |
| *** devanand1 is now known as devananda | 14:11 | |
| devananda | let's try that again | 14:11 |
| * devananda waves | 14:11 | |
| devananda | i'm only slightly here today. mostly in meetings / travelling | 14:11 |
| devananda | lifeless: being 'opinion' surprises me, too. correcting that | 14:12 |
| lucasagomes | devananda, morning | 14:12 |
| devananda | lifeless: it's really not-a-bug since it's not possible for ironic to send an empty pw to ipmitool | 14:13 |
| devananda | lucasagomes: g'morning | 14:13 |
| lucasagomes | devananda, as ur not staying for longer here today, lemme ask you about https://review.openstack.org/#/c/53627/6/ironic/tests/api/test_chassis.py, that is_older_than() test doesn't seems correctly to me, do you think that only checking if the field (created_at/update_at) is populated would be better? | 14:15 |
| * lucasagomes have a cool in 1 min, it's a short one | 14:15 | |
| lucasagomes | call* | 14:15 |
| devananda | that looks a little odd to me, too | 14:16 |
| devananda | i would either check that field is not null OR use timeutils to set a known time, then test to isee if that came back | 14:16 |
| devananda | second seems better | 14:16 |
| *** matsuhas_ has quit IRC | 14:25 | |
| *** matsuhashi has joined #openstack-ironic | 14:25 | |
| *** matsuhashi has quit IRC | 14:30 | |
| NobodyCam | good mornign Ironic | 14:54 |
| lucasagomes | morning NobodyCam | 15:15 |
| lucasagomes | devananda, yea, thanks :) | 15:15 |
| NobodyCam | morning lucasagomes | 15:16 |
| romcheg | Morning guys | 15:21 |
| *** kobier has joined #openstack-ironic | 15:22 | |
| *** jistr is now known as jistr|afk | 15:23 | |
| NobodyCam | lucasagomes: have you seen GheRivero's comments on 54430? | 15:24 |
| NobodyCam | morning romcheg (did see you pop in) | 15:24 |
| lucasagomes | NobodyCam, haven't | 15:24 |
| lucasagomes | but that's the same problem we had in ironic | 15:25 |
| lucasagomes | so | 15:25 |
| lucasagomes | those changes were made on the stable/havanna branch | 15:25 |
| lucasagomes | but the CI script is leaking and submitting those changes to our master | 15:25 |
| lucasagomes | to _all_ the other projects masters as well | 15:25 |
| NobodyCam | :) ahh | 15:25 |
| lucasagomes | https://bugs.launchpad.net/openstack-ci/+bug/1246249 | 15:26 |
| lucasagomes | NobodyCam, thanks for pointing that, I didn't see this one on the client | 15:27 |
| lucasagomes | but yea it's an error | 15:27 |
| lucasagomes | NobodyCam, would be nice to add that gate-ironicclient-requirements, do you know how-to? | 15:28 |
| NobodyCam | lucasagomes: I know That if I asked clarkb he would know! | 15:32 |
| lucasagomes | :D | 15:32 |
| NobodyCam | lucasagomes: in the commit message on 53683 supported = enabled in the conductor, | 15:35 |
| lucasagomes | NobodyCam, hmm yea or active | 15:38 |
| lucasagomes | i think i put supported because of: https://github.com/openstack/ironic/blob/master/ironic/db/api.py#L326 | 15:38 |
| lucasagomes | but I can change | 15:39 |
| NobodyCam | na | 15:39 |
| NobodyCam | just had the thought what happens when we have >1 conductor | 15:39 |
| NobodyCam | and htey have diffenrent drivers | 15:40 |
| lucasagomes | oh yea, that driver list is across all conductors | 15:41 |
| lucasagomes | so it's not actually checking a specific one | 15:41 |
| romcheg | openstack infra is so complicated | 15:43 |
| NobodyCam | :) | 15:44 |
| romcheg | To turn on tests for Ironic I had to patch 2 projects in 3 places | 15:44 |
| lucasagomes | romcheg, haha yea I'm pretty newbie on the infra part | 15:45 |
| NobodyCam | romcheg: do we have a ironic gate job now? | 15:45 |
| romcheg | Not yet | 15:45 |
| romcheg | The patches are published | 15:45 |
| romcheg | but I have to make another one | 15:45 |
| NobodyCam | romcheg: can add the patches to the agenda so others can see them easily | 15:46 |
| *** yuriyz has joined #openstack-ironic | 15:46 | |
| romcheg | Will do that | 15:48 |
| NobodyCam | :) | 15:48 |
| *** blamar has joined #openstack-ironic | 15:49 | |
| NobodyCam | Awesome TY romcheg | 15:49 |
| lucasagomes | awesome2 :) thanks | 15:50 |
| *** rloo has quit IRC | 15:52 | |
| *** rloo has joined #openstack-ironic | 15:52 | |
| romcheg | https://review.openstack.org/#/c/54569/ | 15:53 |
| romcheg | That should fix the problem with the other patch to devstack-gate and after that I will have just 4 published patches that will need to be merged in order to enable devstack tests for ironic :) | 15:54 |
| lucasagomes | haha | 15:54 |
| lucasagomes | sounds over-complicated indeed | 15:55 |
| romcheg | NobodyCam: done with the wiki | 15:59 |
| romcheg | Gotta rush now | 16:00 |
| romcheg | Have a Spanish lesson in 20 minutes | 16:00 |
| NobodyCam | awesome TY romcheg | 16:00 |
| *** lnxnut has joined #openstack-ironic | 16:00 | |
| *** romcheg has quit IRC | 16:03 | |
| *** jistr|afk is now known as jistr | 16:13 | |
| *** romcheg has joined #openstack-ironic | 16:17 | |
| *** Haomeng has quit IRC | 16:22 | |
| *** Haomeng has joined #openstack-ironic | 16:22 | |
| NobodyCam | good $LOCAL_TIME_OF_DAY Haomeng | 16:27 |
| NobodyCam | bbt.... brb | 16:33 |
| *** bauzas has quit IRC | 16:46 | |
| *** jistr has quit IRC | 17:21 | |
| *** harlowja has joined #openstack-ironic | 17:29 | |
| *** yuriyz has quit IRC | 17:33 | |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Required fields on nodes https://review.openstack.org/53664 | 17:36 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add driver interfaces validation https://review.openstack.org/54593 | 17:37 |
| * NobodyCam looks for man page updates | 17:38 | |
| NobodyCam | lol ...JK | 17:38 |
| *** derekh has quit IRC | 17:41 | |
| openstackgerrit | A change was merged to openstack/ironic: Fix test node manager https://review.openstack.org/53877 | 17:46 |
| lucasagomes | NobodyCam, lol | 17:47 |
| lucasagomes | urgh, yea, I'm so lazy to write haha I have it more complete here but I want to put more things on it | 17:47 |
| lucasagomes | like I saw that some manpages they have a section for examples | 17:47 |
| lucasagomes | that would be nice | 17:47 |
| NobodyCam | :) that would be awesome | 17:48 |
| NobodyCam | running to starbucks. brb 15 min | 17:49 |
| NobodyCam | (depending on drive thru line0 | 17:49 |
| lucasagomes | :) | 17:54 |
| jbjohnso | devananda, fyi, I can't really do bridged ipmi comfortably until I have an implementation of such a thing to poke in pyghmi | 17:55 |
| jbjohnso | I can take a good guess | 17:56 |
| jbjohnso | one thing I'd be wondering *hard* is must the user enter so many details or can some of it be picked up on the way | 17:56 |
| jbjohnso | there are a lots of areas where ipmitool pesters the users for stuff that could be autodetected | 17:57 |
| * NobodyCam is back | 18:02 | |
| NobodyCam | jbjohnso: I love autodetect so long as it can be overwritten when it detects things incorrectly for the task I am trying | 18:03 |
| jbjohnso | NobodyCam, well, for example | 18:04 |
| jbjohnso | NobodyCam, using a scheme to navigate RAKP where you need to know whether it is intel or other and how old the firmware is | 18:04 |
| jbjohnso | NobodyCam, when there is a subtle difference that sidesteps that whole mess | 18:04 |
| jbjohnso | NobodyCam, or asking user to provide some obscure bridge address when the intermediate hop offers up the value as a matter of course | 18:05 |
| jbjohnso | a lot of the intelplus and sol hacks in ipmitool are the result of quick one-off patches on top of each other interacting in weird ways | 18:06 |
| jbjohnso | for example, they did get device id but have no good way in their code to segregate SOL and non-SOL at a low level | 18:06 |
| jbjohnso | so they started patching in high level functions all over the place to sidestep | 18:07 |
| jbjohnso | and then someone added a replacement keepalive that wasn't ipmi command | 18:07 |
| jbjohnso | but then it turns out that many ipmi devices don't actually treat empty sol packets as something to keep a connection alive | 18:07 |
| NobodyCam | oh nice | 18:08 |
| jbjohnso | so they made it configurable even though the empty sol packet was a workaround for a bug that I sent a patch in for | 18:08 |
| jbjohnso | of course, me and someone else sent two subtly different fixes for the issue | 18:08 |
| jbjohnso | and they both got merged | 18:08 |
| NobodyCam | ya | 18:08 |
| jbjohnso | and then keepalive *never* detected a broken connection | 18:08 |
| jbjohnso | because our changes had some interesting deadlock when both applied | 18:08 |
| lifeless | devananda: cool | 18:08 |
| NobodyCam | morning lifeless :) | 18:09 |
| jbjohnso | bridging is going to make my route_ipmi_response a tad more complicated, but not too bad | 18:10 |
| jbjohnso | though I wonder how SOL works to those servers | 18:10 |
| jbjohnso | I guess it depends on if the ipmi 2 payloads work in that context... | 18:11 |
| NobodyCam | I need to look at sol more | 18:11 |
| jbjohnso | it might be straightforward... | 18:11 |
| jbjohnso | sol, it ain't pretty, but it's 'good enough' for short haul serial traffic | 18:12 |
| jbjohnso | i.e. ~115200 at most | 18:12 |
| jbjohnso | sol over long haul gets noticably... chunky | 18:12 |
| jbjohnso | for reasons anlogous to why tftp falls apart over long haul, tcp is just so much better at it than what the higher protocls can do | 18:13 |
| jbjohnso | well, not *cn*, but had the will to do (whole point is something simpler than TCP, else they would have just done tcp) | 18:13 |
| NobodyCam | does it support ebcdic | 18:14 |
| jbjohnso | of course it can ;) | 18:15 |
| NobodyCam | :) | 18:15 |
| jbjohnso | SOL is encoding agnostic | 18:15 |
| jbjohnso | it just has data | 18:16 |
| jbjohnso | in fact my web interface | 18:16 |
| jbjohnso | tries multiple encodings | 18:16 |
| jbjohnso | it tries utf-8 then cp452 if not valid utf-8 | 18:16 |
| jbjohnso | because the encoding will actually change over the boot process | 18:16 |
| jbjohnso | BIOS tends to like to do cp452, OS tends to like to do utf-8 | 18:16 |
| jbjohnso | err 437 | 18:17 |
| jbjohnso | in fact, pyghmi might more properly return bytearray instead of string | 18:18 |
| jbjohnso | but meh | 18:18 |
| jbjohnso | bytearray just gives you mutable which isn't relevant for this stuff | 18:18 |
| *** romcheg has quit IRC | 18:31 | |
| *** romcheg has joined #openstack-ironic | 18:33 | |
| NobodyCam | lucasagomes: still awake | 18:43 |
| lucasagomes | NobodyCam, yea, had a call just now | 18:43 |
| lucasagomes | but I'm leaving very soon | 18:43 |
| NobodyCam | :) | 18:43 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add driver interfaces validation https://review.openstack.org/54593 | 18:43 |
| lucasagomes | NobodyCam, another thing | 18:43 |
| lucasagomes | I was looking the git manpage | 18:43 |
| lucasagomes | and they break their commands into sections | 18:43 |
| lucasagomes | I was thinking about having like node, port, chassis sections for us as well | 18:44 |
| lucasagomes | what you think? | 18:44 |
| lucasagomes | instead of having everything in one man page | 18:44 |
| lucasagomes | NobodyCam, like https://www.kernel.org/pub/software/scm/git/docs/git.html | 18:44 |
| lucasagomes | https://www.kernel.org/pub/software/scm/git/docs/git-add.html | 18:44 |
| NobodyCam | ++ | 18:45 |
| NobodyCam | ya I like that | 18:45 |
| lucasagomes | yea I will look some other manpages as examples | 18:45 |
| lucasagomes | they r all diff man | 18:45 |
| lucasagomes | I will try to come up with something nice for us | 18:46 |
| NobodyCam | just some to think about | 18:46 |
| NobodyCam | # NOTE(d): For simpler code paths on the deployment side, | 18:46 |
| NobodyCam | # we always create a swap partition. If the flavor | 18:46 |
| NobodyCam | # does not specify any swap, we default to 1MB | 18:46 |
| NobodyCam | thats from 51328 | 18:47 |
| NobodyCam | what do you think about making a valid swap size ie. == ram | 18:47 |
| devananda | -- | 18:48 |
| lucasagomes | NobodyCam, well we don't know if there's space for that | 18:48 |
| NobodyCam | :) hi devananda | 18:48 |
| devananda | hi! | 18:48 |
| lucasagomes | ram could be like 16gb at least so might need some space | 18:48 |
| lucasagomes | although, at first glance having 1mb swap also looks odd | 18:49 |
| NobodyCam | ya | 18:49 |
| lucasagomes | I mean, it's pretty much useless | 18:49 |
| NobodyCam | lol | 18:49 |
| romcheg | devananda: Morning | 18:50 |
| romcheg | Can you please ping someone to look at https://review.openstack.org/#/c/54569/ ? | 18:50 |
| NobodyCam | if we ever look at putting system to sleep will need a valid swap size | 18:51 |
| NobodyCam | thou we could create a swap file from with in the os itself and do a swapon kinda thing | 18:52 |
| openstackgerrit | Lucas Alvares Gomes proposed a change to openstack/ironic: Add driver interfaces validation https://review.openstack.org/54593 | 18:52 |
| lucasagomes | NobodyCam, yea could use a file on the fs | 18:52 |
| lucasagomes | but hmm, I have to think more about it | 18:53 |
| NobodyCam | devananda: how close are the nova bindings | 18:53 |
| lucasagomes | another thing I think that would be cool is a way to use 100% of the disk | 18:53 |
| NobodyCam | I wanta do a nova-ironic element | 18:53 |
| lucasagomes | for example if root_gb is not specified you could just use evertyhing (- swap size if specified) | 18:53 |
| NobodyCam | lucasagomes: I not so fond of that. | 18:54 |
| NobodyCam | admins may want empty space for snap shots | 18:54 |
| NobodyCam | and other such things | 18:55 |
| lucasagomes | well that is optional, whether he wants to use all the space or not | 18:55 |
| NobodyCam | I like use what root_gb says to use. Now something like root_gb=-1 | 18:55 |
| NobodyCam | = use all space | 18:56 |
| lucasagomes | I mean making it optional he doesn't even know exactly how much space there's on that disk he just says "use it all" | 18:56 |
| lucasagomes | NobodyCam, oh yea | 18:56 |
| lucasagomes | what I mean is, if root_gb is not specified or as u said, root_gb=-1 (as a flag) | 18:57 |
| devananda | NobodyCam: not close. lemme get the review | 18:57 |
| NobodyCam | thats also use to match flavor to node in sch | 18:57 |
| devananda | https://review.openstack.org/#/c/51328/ | 18:57 |
| lucasagomes | yea :( the way the scheduler finds a node is not very baremetal-friendly | 18:58 |
| devananda | hm? | 18:58 |
| lucasagomes | cause for baremetal we wouldn't care about the numer of cpus or ram for example, it's always going to use 100% | 18:58 |
| devananda | lucasagomes: scheduler should do flavor == node spec | 18:58 |
| devananda | it may not right now .. but it should :) | 18:58 |
| NobodyCam | lucasagomes: we care for diff hardware specs | 18:58 |
| devananda | also, ironi has to expose all nodes to nova-scheduler. which doesn't scale when there are 100k nodes | 18:59 |
| devananda | so host aggregaces or AZs will be needed | 18:59 |
| devananda | but i digress | 18:59 |
| lucasagomes | :) I see | 18:59 |
| lifeless | 100K in one scheduler should be fine | 19:00 |
| lifeless | isn't geekinutah running 400K hypervisors in one scheduler ? | 19:00 |
| lifeless | NobodyCam: huh, root_gb=-1 isn't a meaningful flavor | 19:00 |
| lifeless | NobodyCam: flavors define the characteristics of the machine to users | 19:00 |
| NobodyCam | lifeless: yes | 19:01 |
| lifeless | NobodyCam: for local stuff, thats what ephemeral can be used for | 19:01 |
| NobodyCam | just talking with lucas about that | 19:01 |
| lucasagomes | :) | 19:02 |
| devananda | lifeless: i thought that was 40k | 19:03 |
| devananda | or maybe i missed some conversation | 19:03 |
| lifeless | devananda: nope | 19:03 |
| devananda | wow. then, that's awesome | 19:03 |
| lifeless | devananda: ah, no 400K vms | 19:04 |
| devananda | NobodyCam: if you were looking to push teh nova driver along, feel free to hack on and upload new revs for https://review.openstack.org/#/c/51328/ | 19:04 |
| devananda | lifeless: ahh. that makes more sense. | 19:04 |
| NobodyCam | :) | 19:05 |
| lifeless | devananda: I'm just digging for chapter and verse | 19:05 |
| lifeless | nope, what I found was a different thing again | 19:08 |
| lifeless | I dunno | 19:08 |
| lifeless | 400K rows to consult in python is fairly small biscuits though | 19:08 |
| lifeless | particularly if indexed well | 19:08 |
| *** lucasagomes is now known as lucas-dinner | 19:11 | |
| *** ndipanov is now known as ndipanov_gone | 19:11 | |
| devananda | lifeless: it's the "fetch 400k rows from db, iterate in nested loops" that bothers me. but maybe that's been improved in havana | 19:12 |
| *** martyntaylor has quit IRC | 19:40 | |
| *** martyntaylor1 has joined #openstack-ironic | 19:40 | |
| linggao | Hi lifeless, is HasLength a python function? | 19:44 |
| linggao | I googled it and cannot find it. | 19:44 |
| lifeless | linggao: it's in testtools.matchers - from testtools.matchers import HasLength #noqa | 19:46 |
| linggao | self.assertThat([1, 2, 3], HasLength(2)) | 19:49 |
| linggao | ? | 19:49 |
| lifeless | for instance, yes. | 19:51 |
| lifeless | assertThat(subject, matcher) | 19:51 |
| lifeless | assert that [1,2,3] has length 2 | 19:51 |
| lifeless | devananda: yeah, I agree work may be needed. | 19:52 |
| lifeless | devananda: I don't think we should design around making nova better though :) | 19:52 |
| linggao | And lifeless, another thing is that you are saying that you prefer the expected value on left hand side for assert. | 19:52 |
| linggao | Is this a standard for openstack or just personal preference? I saw a lot of places the expected values are on the right hand side. | 19:53 |
| linggao | I personally like to put it on right hand side. | 19:53 |
| lifeless | testtools.TestCase.assertEqual = assertEqual(self, expected, observed, message='') unbound testtools.testcase.TestCase method | 19:54 |
| lifeless | Assert that 'expected' is equal to 'observed'. | 19:54 |
| lifeless | 19:54 | |
| lifeless | :param expected: The expected value. | 19:54 |
| lifeless | :param observed: The observed value. | 19:54 |
| lifeless | :param message: An optional message to include in the error. | 19:54 |
| lifeless | linggao: it is neither openstack specific nor personal preference. | 19:54 |
| lifeless | linggao: it is how the function is defined. | 19:54 |
| lifeless | linggao: the left side/right side thing is obviously a minor nit, and I wouldn't -1 a patch that had only that wrong with it. | 19:56 |
| lifeless | linggao: but when there are other things to fix up, I will mention it. | 19:56 |
| linggao | lifeless, I understand. I just want to make sure that if I miss anything. | 20:04 |
| linggao | For example: | 20:05 |
| linggao | assertEqual(first, second, msg=None) | 20:05 |
| linggao | Test that first and second are equal. If the values do not compare equal, the test will fail. | 20:05 |
| linggao | This is what I got from the python doc http://docs.python.org/3/library/unittest.html | 20:05 |
| lifeless | linggao: try 'pydoc testtools.TestCase.assertEqual' | 20:05 |
| linggao | ok, got it. thanks a lot. | 20:06 |
| linggao | for the important parts of the review, the API and the CLI are discussed here already with devananda and NobodyCam and other. You can bring them up if you have questions. | 20:07 |
| linggao | I can change the client API to be separate function though if the parameters are confusing. | 20:08 |
| linggao | lifeless, I mean you can bring them up here so we can discuss them again together. | 20:09 |
| lifeless | linggao: sure, this is the instance_uuid api right ? | 20:13 |
| linggao | lifeless, yes | 20:14 |
| linggao | \/nodes/?instance_uuid=<uuid|null|notnull> | 20:14 |
| lifeless | righto | 20:15 |
| lifeless | linggao: got a link to the review? | 20:15 |
| linggao | https://review.openstack.org/#/c/53485/6/ironicclient/tests/v1/test_node.py | 20:16 |
| linggao | Linr 140 | 20:16 |
| linggao | Line | 20:16 |
| lifeless | devananda: hey so - in this review | 20:18 |
| lifeless | devananda: I had some detailed feedback, but there is a broader question about the CLI | 20:18 |
| lifeless | devananda: which linggao would like us to reach consensus on rather than being tossed around by reviewer comments :P | 20:19 |
| linggao | :) | 20:19 |
| lifeless | devananda: I am suggesting that 'node-show' should stay super precise, and that -list is where all the instance_uuid querying should take place | 20:19 |
| lifeless | devananda: I realise you're in meetings right now; I have to pop out to get my hair done so I don't impersonate a shaggy dog @ HK | 20:20 |
| lifeless | devananda: but we should be able to carry on a moderate latency chat here? | 20:20 |
| *** martyntaylor1 has left #openstack-ironic | 20:22 | |
| * NobodyCam notes that he has now added hair cut to his list of things that need doing before HK | 20:24 | |
| linggao | lol | 20:24 |
| rloo | NobodyCam: I managed to get tripleo+ironic working this morning. I'll try to update the walkthru today, to use neutron-dhcp-agent. | 20:25 |
| NobodyCam | linggao: I will be updating the walkthru based on https://review.openstack.org/#/c/44500/ | 20:26 |
| lifeless | ok, -> hair dressers bbiaw | 20:26 |
| rloo | NobodyCam: then I won't touch the walkthru? | 20:27 |
| NobodyCam | :) | 20:28 |
| rloo | quick question (I hope). I am adding a 'last_error' field to nodes db table. How do I test migration? | 20:28 |
| linggao | NobodayCam, I do not know what is that for. | 20:29 |
| rloo | linggao: I think he meant me :-) | 20:30 |
| linggao | ah. | 20:30 |
| NobodyCam | :-p | 20:30 |
| NobodyCam | y yes I did | 20:30 |
| devananda | lifeless: yep, we can have async chat. i'll haev irc access for next few hours, then dinner & flight home | 20:30 |
| devananda | lifeless: a uuid is not a free-form string. it is (or should be) checked against uuidutils.is_uuid_like() | 20:32 |
| devananda | lifeless: null and notnull are not valid uuids | 20:33 |
| linggao | davananda, lifeless: here is current implementation. the url is .../ nodes/?instance_uuid=<uuid|null|notnull> | 20:35 |
| linggao | The added CLIs are: node-show [--instance-uuid] <id> | 20:38 |
| linggao | node-list [--associated|--unassociated\ | 20:39 |
| NobodyCam | node-list [--[un]associated] :-p | 20:40 |
| linggao | NobodyCam, yes. | 20:41 |
| linggao | lifeless is gone for hair cur :) | 20:42 |
| devananda | linggao: I am comfortable with both that API and CLI. it seems lifeless is not, and i'm curious to know what he proposes for those | 20:44 |
| devananda | linggao: there are also several comments from lifeless on the review which I agree with, regarding the internal API. | 20:44 |
| devananda | for example, def list(self, associated=True, unassociated=True) -- I don't ilke this either | 20:45 |
| NobodyCam | devananda: have you seen lifeless's comments on https://review.openstack.org/#/c/53485/6/ironicclient/tests/v1/test_node.py | 20:45 |
| devananda | yes | 20:46 |
| NobodyCam | :) | 20:46 |
| linggao | devananda, for the internal API, I'll make them separate functions. | 20:47 |
| devananda | linggao: see my comments above | 20:47 |
| linggao | dacananda, for def list(self, associated=True, unassociated=True) | 20:51 |
| linggao | Can I make them 3 different APIs? | 20:52 |
| linggao | get, get_associated, get_unassociated? | 20:52 |
| linggao | sorry get/list | 20:52 |
| linggao | devananda, for def get(self, node_id, instance_uuid=False): I will split them to get and get_by_instance_uuid. All these are done in my sandbox. | 20:54 |
| linggao | My original attempt was not adding too may APIs but adding a few parameters to existing APIs. If this is confusing, then separate APIs seems a solution. | 20:55 |
| devananda | lifeless: suggestion is to use one parameter, overloaded with 4 possible options | 20:56 |
| devananda | <uuid>, None, True, False | 20:56 |
| linggao | that will be more confusing. | 20:57 |
| devananda | how do you propose to expose your 3 methods in the REST API ? | 20:58 |
| linggao | nodes/?instance_uuid=<uuid|null|notnull> | 20:58 |
| devananda | so, same | 20:59 |
| linggao | so with lifeless's suggestion, node-show will not show by instance_uuid, right? | 21:00 |
| linggao | It is moved to node-list? | 21:00 |
| devananda | that is confusing to me | 21:01 |
| NobodyCam | i find it less confusing. | 21:02 |
| devananda | node-list should return a list of uuids | 21:02 |
| linggao | node-show shows one node with more details. | 21:02 |
| devananda | node-show should return the details of a node | 21:02 |
| linggao | node-list shows a collection of nodes. | 21:02 |
| *** jbjohnso has quit IRC | 21:02 | |
| devananda | right | 21:02 |
| devananda | so it would be very inconsistent for node-list to return a single node with full details expanded | 21:03 |
| linggao | yes. | 21:03 |
| NobodyCam | filter the list to get what nodes you want then show those nodes by filtered list returned | 21:03 |
| NobodyCam | asking node show for all associated nodes seems inconsistent to me | 21:05 |
| *** anniec has joined #openstack-ironic | 21:05 | |
| NobodyCam | [un]associated | 21:05 |
| linggao | NobodyCam, node-show only shows one node. node-show --instance-uuid <id> | 21:06 |
| linggao | It has the same detailed output as node-show <nodeid> | 21:07 |
| NobodyCam | what would null and not null give | 21:07 |
| linggao | that's in node-list | 21:07 |
| linggao | node-list [--[un]associated] | 21:08 |
| NobodyCam | yes | 21:08 |
| NobodyCam | your example was nodes/?instance_uuid=<uuid|null|notnull> | 21:08 |
| linggao | node-show will call internal API get, node-list call list. | 21:08 |
| NobodyCam | seems /me is confused about what bits are confusing whom | 21:14 |
| linggao | davananda, lifeless, I do not understand neither of your comments. on https://review.openstack.org/#/c/53485/6/ironicclient/tests/v1/test_node.py Line 140. Is there something I need to change there? | 21:19 |
| linggao | are you debating with each other or telling me the same thing? :) | 21:20 |
| NobodyCam | linggao: they are working that out... thus the async convo going on atm | 21:21 |
| linggao | NobodyCam, I see. | 21:22 |
| *** jdob has quit IRC | 21:40 | |
| *** romcheg has left #openstack-ironic | 21:56 | |
| *** linggao has quit IRC | 22:02 | |
| *** anniec has quit IRC | 22:05 | |
| *** rloo has quit IRC | 22:06 | |
| *** rloo has joined #openstack-ironic | 22:06 | |
| *** anniec has joined #openstack-ironic | 22:07 | |
| *** anniec has quit IRC | 22:12 | |
| *** linggao has joined #openstack-ironic | 22:26 | |
| *** crandquist has joined #openstack-ironic | 22:53 | |
| *** kobier has quit IRC | 22:56 | |
| * NobodyCam wanders away | 23:33 | |
| *** rwsu has quit IRC | 23:38 | |
| *** rwsu has joined #openstack-ironic | 23:45 | |
| *** lucas-dinner has quit IRC | 23:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!