| *** jmankov has joined #openstack-ironic | 00:04 | |
| NobodyCam | three? | 00:04 |
|---|---|---|
| jroll | seven | 00:05 |
| *** jmanko has quit IRC | 00:07 | |
| NobodyCam | clif_h: https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/tests/agent.py#L333 | 00:09 |
| *** Masahiro has joined #openstack-ironic | 00:10 | |
| *** Masahiro has quit IRC | 00:14 | |
| *** spandhe has left #openstack-ironic | 00:14 | |
| *** openstackgerrit has quit IRC | 00:18 | |
| *** openstackgerrit has joined #openstack-ironic | 00:19 | |
| *** andreykurilin_ has quit IRC | 00:21 | |
| *** spandhe_ has joined #openstack-ironic | 00:25 | |
| JayF | clif_h: ^ I think NobodyCam just found your test failure | 00:25 |
| NobodyCam | :-p | 00:26 |
| clif_h | NobodyCam: Thank you | 00:26 |
| NobodyCam | nobodycamAir:ironic-python-agent NobodyCam$ grep -R _get_kernel_params * :-p | 00:26 |
| clif_h | I thought I had grepped through the repo for that function name | 00:27 |
| clif_h | but I did not hit the tests | 00:27 |
| clif_h | :( | 00:27 |
| NobodyCam | :) | 00:27 |
| JayF | if you get a merge in 2 patchsets | 00:27 |
| JayF | for your first openstack contribution | 00:27 |
| *** spandhe_ has left #openstack-ironic | 00:28 | |
| JayF | That's like, 96 less than JoshNang needed to not get his original agent driver merged :P | 00:28 |
| NobodyCam | lol | 00:28 |
| NobodyCam | JayF: you didn't say be nice to JoshNang the first day | 00:28 |
| NobodyCam | lol | 00:29 |
| * NobodyCam *ducks* | 00:29 | |
| clif_h | im sure the agent driver was much more complex though | 00:29 |
| JayF | You guys wouldn't have listened to me anyway :) | 00:29 |
| clif_h | this is touching-the-ground-fruit | 00:29 |
| NobodyCam | hehehehe :-p | 00:29 |
| JayF | clif_h: you have no idea ... we basically had upstreamed, in a single gerrit patchset, all our downstream code. It was giant and awful and unreviewable. | 00:29 |
| NobodyCam | +2000 lines | 00:30 |
| NobodyCam | but it really was a "new" approch (and quite kewl) | 00:30 |
| JayF | it was worse than that, wasn't it? | 00:30 |
| JayF | NobodyCam: we ran it in production for a long time (still kinda do; but it's broken up even downstream now) | 00:31 |
| *** marcoemorais has quit IRC | 00:31 | |
| *** marcoemorais has joined #openstack-ironic | 00:31 | |
| JayF | clif_h: the important thing is the tempest job passed; that means you didn't break anything but tests :D | 00:33 |
| *** marcoemorais has quit IRC | 00:33 | |
| *** marcoemorais has joined #openstack-ironic | 00:33 | |
| *** hemna has quit IRC | 00:34 | |
| openstackgerrit | Clif Houck proposed openstack/ironic-python-agent: Fix badly named function _get_kernel_params() https://review.openstack.org/139269 | 00:35 |
| clif_h | JayF: but all the tests must pass! | 00:36 |
| clif_h | :) | 00:36 |
| *** penick has joined #openstack-ironic | 00:38 | |
| jroll | clif_h: https://github.com/ggreer/the_silver_searcher | 00:39 |
| *** Haomeng|2 has quit IRC | 00:45 | |
| *** Haomeng|2 has joined #openstack-ironic | 00:47 | |
| *** shakamunyi has quit IRC | 00:50 | |
| *** spandhe has joined #openstack-ironic | 00:51 | |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Refactor async helper methods in conductor/manager.py https://review.openstack.org/139217 | 00:52 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Begin using the state machine for node deploy/teardown https://review.openstack.org/139216 | 00:52 |
| *** igordcard has quit IRC | 00:52 | |
| *** penick has quit IRC | 01:04 | |
| *** Masahiro has joined #openstack-ironic | 01:06 | |
| *** dlaube has quit IRC | 01:09 | |
| *** penick has joined #openstack-ironic | 01:11 | |
| *** penick has quit IRC | 01:13 | |
| openstackgerrit | Merged openstack/ironic-python-agent: Fix badly named function _get_kernel_params() https://review.openstack.org/139269 | 01:15 |
| JayF | clif_h: ^ \o/ grats | 01:16 |
| *** jerryz has joined #openstack-ironic | 01:17 | |
| openstackgerrit | Merged openstack/ironic: Updated from global requirements https://review.openstack.org/139229 | 01:18 |
| clif_h | JayF: \o/ thanks! | 01:28 |
| NobodyCam | congratz clif_h | 01:30 |
| JayF | NobodyCam: if you know of any good bugs or small things to point clif_h at, please reccomend them. our low-hanging-fruit tagged bugs are more like rotten fruit that's been on the ground for a year | 01:32 |
| clif_h | NobodyCam: Thanks! | 01:32 |
| JayF | (aka all docs updates or logging improvements) | 01:32 |
| devananda | one of the things I'd love to see is a good operator-centric logging rewrite within the ConductorManager module | 01:33 |
| jroll | we don't know nothin' about operating ironic | 01:34 |
| * jroll runs | 01:34 | |
| devananda | heh heh heh | 01:34 |
| JayF | http://www.stopbuyingcrap.com/pics/sbc/itsatrap.jpg | 01:34 |
| devananda | JayF: you walked into that one | 01:34 |
| devananda | (I'm just sneaky, so you didn't see it coming) | 01:35 |
| devananda | actually, I'm also frustrated by it after jumping in that code again today | 01:35 |
| JayF | Well if I was going to fix a log message, I'd fix the case jroll found the other day where ipa can fail to download an image and not log a damn thing | 01:36 |
| JayF | which seems ... bad | 01:36 |
| devananda | that doesn't seem helpful, no | 01:36 |
| JayF | but we knew what happened because we had logs in devstack for ipa | 01:36 |
| JayF | so progress at least | 01:36 |
| jroll | I mean, ipa logging story is horrible | 01:36 |
| jroll | so is the docs story | 01:36 |
| jroll | both make me feel like a bad person | 01:36 |
| JayF | devananda: https://review.openstack.org/#/c/134436/ itsatrap.jpg | 01:36 |
| jroll | (I might be) | 01:36 |
| devananda | if I was refactoring ConductorManager, I might do this: https://review.openstack.org/#/c/139217/2/ironic/conductor/manager.py | 01:36 |
| devananda | :) | 01:37 |
| JayF | does ironic-parallel tempest usually pass nowadays? | 01:37 |
| devananda | that'sa good question | 01:37 |
| devananda | I see several cases where it failed but pxe_ssh passed | 01:38 |
| JayF | Yeah, I should just check adam_g's gate status thing | 01:39 |
| adam_g | JayF, it should be | 01:39 |
| adam_g | wel | 01:39 |
| adam_g | theres that race thing we were looking at the other week that turned up there as well | 01:39 |
| JayF | does it happen more often/more likely in the parallel job, I wonder? That could be an interesting data point | 01:40 |
| adam_g | JayF, its more likely to happen the parallel job because we're launching so many more VMs there than the current pxe_ssh job | 01:40 |
| JayF | adam_g: did you make any progress on the persistent logs from libvirt console? I sadly never got around to testing that locally | 01:40 |
| JayF | adam_g: that's what I thought, yeah | 01:40 |
| adam_g | JayF, but the failure started to pop up more often in the pxe_ssh job, but seems to have calmed down, mysteriously | 01:40 |
| adam_g | JayF, i actually haven't had a chance on that, sorry :| | 01:41 |
| JayF | It's fine; I haven't tested it locally so I can't complain | 01:41 |
| JayF | heh | 01:41 |
| *** vdrok has quit IRC | 01:41 | |
| adam_g | JayF, i was hoping the current patchset @ https://review.openstack.org/#/c/129099/ would work, it worked locally when i spawned locally but doing it in the slave didnt work out so well | 01:42 |
| *** vdrok has joined #openstack-ironic | 01:42 | |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Refactor async helper methods in conductor/manager.py https://review.openstack.org/139217 | 01:43 |
| openstackgerrit | Devananda van der Veen proposed openstack/ironic: Begin using the state machine for node deploy/teardown https://review.openstack.org/139216 | 01:43 |
| devananda | anyone else seeing this in the conductor logs lately? | 01:43 |
| devananda | DEBUG oslo.messaging._drivers.impl_rabbit [-] Timed out waiting for RPC response: timed out _raise_timeout_if_deadline_is_reached | 01:43 |
| devananda | adam_g: waht's the race you're seeing? | 01:45 |
| adam_g | devananda, yeah i just noticed those somewhere this past hour | 01:45 |
| adam_g | devananda, some issue where a node sporadically cannot reach back to the host machines tftp server. not certain its a race or what | 01:46 |
| devananda | ok | 01:46 |
| devananda | the failures I've got right now in parallel-nv are u'No valid host was found. There are not enough hosts available.' | 01:46 |
| *** smoriya has joined #openstack-ironic | 01:47 | |
| devananda | seems to be on this patch set only -- https://review.openstack.org/#/c/139217/ | 01:47 |
| devananda | if this news to you, I'll assume it's something I broke :) | 01:47 |
| adam_g | devananda, well i was seeing these earlier, but patched devstack to address it | 01:48 |
| adam_g | (it was starting tempest tests before n-cpu's periodic picked up recently enrolled nodes) | 01:49 |
| adam_g | im actually looking at the same failure on 139223 | 01:49 |
| adam_g | which are certainly not related to the patch | 01:49 |
| devananda | ah, k | 01:50 |
| *** nosnos has joined #openstack-ironic | 01:52 | |
| devananda | adam_g: yea, looks like the same situation here. tests failed before n-cpu picked up available resources | 01:52 |
| *** dlaube1 has joined #openstack-ironic | 01:53 | |
| adam_g | devananda, yeah | 01:53 |
| adam_g | http://logs.openstack.org/23/139223/1/check/check-tempest-dsvm-ironic-parallel-nv/cb2bf22/logs/devstacklog.txt.gz#_2014-12-04_22_51_03_347 | 01:53 |
| adam_g | devstack should wait until 'hypervisor-stats -ge $expected_nodes' | 01:53 |
| adam_g | which its doing, but still failing. | 01:53 |
| adam_g | devananda, re those messaging timeouts, i wonder if those starting popping up around the time new oslo.messaging was released today | 01:54 |
| devananda | adam_g: oh. right. makes sense | 01:59 |
| adam_g | i need to run i wanna look closer at those new scheduler fails tomorrow, timestamps seem to checkout on 139223--the failures happen long after nova's got the resources, i wonder if tempest is spinning up more instances concurrently now | 02:06 |
| *** ChuckC_ has quit IRC | 02:12 | |
| *** ChuckC_ has joined #openstack-ironic | 02:12 | |
| *** dlaube1 has quit IRC | 02:12 | |
| *** alexpilotti has joined #openstack-ironic | 02:15 | |
| *** dlaube has joined #openstack-ironic | 02:17 | |
| *** ChuckC_ has quit IRC | 02:17 | |
| *** kurtrao has quit IRC | 02:22 | |
| *** Masahiro has quit IRC | 02:22 | |
| *** rloo has quit IRC | 02:23 | |
| *** Masahiro has joined #openstack-ironic | 02:24 | |
| *** dlaube has quit IRC | 02:29 | |
| *** kurtrao has joined #openstack-ironic | 02:31 | |
| *** marcoemorais has quit IRC | 02:37 | |
| *** ryanpetrello has quit IRC | 02:40 | |
| *** ramineni has joined #openstack-ironic | 02:41 | |
| *** ryanpetrello has joined #openstack-ironic | 02:43 | |
| *** nosnos has quit IRC | 02:48 | |
| *** ChuckC_ has joined #openstack-ironic | 02:49 | |
| *** r-daneel has quit IRC | 02:50 | |
| *** shakamunyi has joined #openstack-ironic | 02:50 | |
| *** shakamunyi has quit IRC | 02:50 | |
| *** Haomeng has joined #openstack-ironic | 02:51 | |
| *** Haomeng|2 has quit IRC | 02:52 | |
| *** harlowja_ is now known as harlowja_away | 02:54 | |
| *** spandhe has quit IRC | 02:57 | |
| *** shakamunyi has joined #openstack-ironic | 02:59 | |
| *** achanda has joined #openstack-ironic | 03:01 | |
| *** achanda has quit IRC | 03:11 | |
| *** dlaube has joined #openstack-ironic | 03:15 | |
| *** naohirot has quit IRC | 03:26 | |
| *** Masahiro has quit IRC | 03:38 | |
| openstackgerrit | Jeremy Stanley proposed openstack/ironic: Workflow documentation is now in infra-manual https://review.openstack.org/139329 | 03:41 |
| openstackgerrit | Jeremy Stanley proposed openstack/ironic-python-agent: Workflow documentation is now in infra-manual https://review.openstack.org/139330 | 03:41 |
| openstackgerrit | Jeremy Stanley proposed openstack/ironic-specs: Workflow documentation is now in infra-manual https://review.openstack.org/139331 | 03:41 |
| *** Masahiro has joined #openstack-ironic | 03:44 | |
| *** stevebaker has left #openstack-ironic | 03:45 | |
| *** Masahiro has quit IRC | 03:49 | |
| openstackgerrit | Jeremy Stanley proposed openstack/python-ironicclient: Workflow documentation is now in infra-manual https://review.openstack.org/139374 | 03:51 |
| *** pcrews has quit IRC | 03:58 | |
| *** Marga_ has quit IRC | 03:59 | |
| *** Masahiro has joined #openstack-ironic | 03:59 | |
| *** alexpilotti has quit IRC | 04:00 | |
| *** naohirot has joined #openstack-ironic | 04:04 | |
| *** pensu has joined #openstack-ironic | 04:05 | |
| naohirot | good afternoon ironic! | 04:06 |
| mrda | hi naohirot | 04:07 |
| naohirot | mrda: hi | 04:07 |
| * naohirot I had a long meeting this morning :) | 04:07 | |
| *** r-daneel has joined #openstack-ironic | 04:15 | |
| *** Marga_ has joined #openstack-ironic | 04:37 | |
| *** rameshg87 has joined #openstack-ironic | 04:37 | |
| *** dlaube has quit IRC | 04:38 | |
| *** r-daneel has quit IRC | 04:53 | |
| *** killer_prince is now known as lazy_prince | 04:53 | |
| Haomeng | naohirot: hi | 05:01 |
| naohirot | Haomeng: Hi :) | 05:02 |
| Haomeng | naohirot: good afternoon:) | 05:02 |
| naohirot | Haomeng: good afternoon | 05:03 |
| Haomeng | naohirot: :) | 05:03 |
| naohirot | Haomeng: are you going to attend the next IRC meeting which starts 5 UTC? | 05:04 |
| Haomeng | naohirot: sure, I want, my zone is +8, it should be my 1pm? | 05:09 |
| Haomeng | 8 dec, 5utc,right? | 05:09 |
| naohirot | Haomeng: Yes, it is, it should be 1PM, and 2PM here. | 05:10 |
| Haomeng | naohirot: we have two weeks IRC with utc+5, right? | 05:10 |
| naohirot | Haomeng: Yes, but there are good news and bad news. | 05:11 |
| Haomeng | naohirot: :) | 05:11 |
| Haomeng | naohirot: I like bad news:) | 05:11 |
| naohirot | Haomeng: good news is that new time schedule is absolutely easier to attend. | 05:12 |
| *** Masahiro has quit IRC | 05:12 | |
| Haomeng | naohirot: yes | 05:12 |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Cisco UCS power driver https://review.openstack.org/139517 | 05:12 |
| naohirot | Haomeng: Bad news is that it's easier to conflict schedule :< | 05:12 |
| naohirot | Haomeng: So I cannot attend to the next meeting due to double booking. | 05:13 |
| Haomeng | naohirot: last release meeting is in my 3am, I attended three times totally:) | 05:13 |
| Haomeng | naohirot: dont worry | 05:13 |
| *** Marga_ has quit IRC | 05:13 | |
| Haomeng | Haomeng: I have habit that i will review the meeting log if I can not attend | 05:13 |
| Haomeng | naohirot: to catch the points:) | 05:13 |
| Haomeng | naohirot: dont worry | 05:13 |
| *** Marga_ has joined #openstack-ironic | 05:14 | |
| naohirot | Haomeng: Yes, I'll check the log. | 05:14 |
| Haomeng | naohirot: :) | 05:14 |
| Haomeng | naohirot: for kilo meeting, we are runing with new mode, to discuss the topis which is raised before meeting, so we have chance to discuss topics in meeting that is cool | 05:15 |
| naohirot | Haomeng: If starting at 3am our local time never creates double booking :-) | 05:15 |
| Haomeng | naohirot: :) | 05:15 |
| Haomeng | naohirot: np, dont worry | 05:15 |
| naohirot | Haomeng: Okay :-) | 05:16 |
| Haomeng | naohirot: :) | 05:16 |
| *** Masahiro has joined #openstack-ironic | 05:27 | |
| *** saripurigopi has joined #openstack-ironic | 05:52 | |
| saripurigopi | HI, I'm trying to submit new BP for kilo. | 05:54 |
| saripurigopi | I've executed the tox tests locally, and do not see any errors. But Jenkins tests failed with an error. | 05:54 |
| saripurigopi | ft1.2: tests.test_titles.TestTitles.test_current_cycle_template_StringException: Traceback (most recent call last): | 05:55 |
| saripurigopi | File "tests/test_titles.py", line 142, in test_current_cycle_template | 05:55 |
| saripurigopi | self._check_license(data) | 05:55 |
| saripurigopi | self._check_license(data) | 05:55 |
| saripurigopi | any idea what I'm missing here. | 05:55 |
| Haomeng | saripurigopi: hi | 05:56 |
| saripurigopi | @Haomeng Hi | 05:57 |
| Haomeng | saripurigopi: let me help you check | 05:57 |
| Haomeng | saripurigopi: which one? | 05:57 |
| Haomeng | https://review.openstack.org/#/c/139517/? | 05:58 |
| saripurigopi | Haomeng: This is the failure result http://logs.openstack.org/17/139517/1/check/gate-ironic-specs-python27/c6d67f3/testr_results.html.gz | 05:58 |
| Haomeng | saripurigopi: ok | 05:58 |
| saripurigopi | yes | 05:58 |
| *** lazy_prince has quit IRC | 05:58 | |
| Haomeng | saripurigopi: we can check this log in details - http://logs.openstack.org/17/139517/1/check/gate-ironic-specs-python27/c6d67f3/console.html | 05:59 |
| *** penick has joined #openstack-ironic | 05:59 | |
| Haomeng | saripurigopi: from the first error message | 05:59 |
| Haomeng | saripurigopi: missing the license in the top of spec I think | 06:00 |
| saripurigopi | Haomeng: ok, let me check. | 06:00 |
| Haomeng | saripurigopi: I check your license lines, looks fine | 06:01 |
| *** penick has quit IRC | 06:01 | |
| Haomeng | saripurigopi: so dont worry, some times that is caused by jenkins | 06:01 |
| saripurigopi | Haomeng: okay, | 06:01 |
| Haomeng | saripurigopi: and some of the cases are blocked by other code | 06:01 |
| Haomeng | did you notice "Attribution 3.0 Unported\" | 06:02 |
| saripurigopi | Haomeng: yeah | 06:02 |
| Haomeng | saripurigopi: your first line, can you try to remove the last char "\" | 06:02 |
| Haomeng | saripurigopi: maybe it is not required and break the license validator | 06:02 |
| Haomeng | saripurigopi: maybe | 06:02 |
| Haomeng | saripurigopi: :) | 06:02 |
| openstackgerrit | Gopi Krishna S proposed openstack/ironic-specs: Cisco UCS power driver https://review.openstack.org/139517 | 06:04 |
| saripurigopi | Haomeng: submitted again, after removing the last char. | 06:05 |
| Haomeng | saripurigopi: ok, good luck:) | 06:05 |
| Haomeng | saripurigopi: :) | 06:05 |
| *** shakamunyi has quit IRC | 06:07 | |
| Haomeng | saripurigopi: works:) | 06:08 |
| saripurigopi | Haomeng: Yes , thank you :) | 06:09 |
| Haomeng | Jenkins +1 | 06:09 |
| Haomeng | :) | 06:09 |
| Haomeng | welcome | 06:09 |
| saripurigopi | Haomeng: is 8th Dec, the last date to submit new BP? | 06:09 |
| *** rushiagr is now known as rushiagr_away | 06:10 | |
| Haomeng | saripurigopi: should be, let me check the reference | 06:14 |
| Haomeng | saripurigopi: but dont worry, you have raised the spec already | 06:15 |
| saripurigopi | Haomeng: okay, thank you. | 06:18 |
| *** mrda is now known as mrda-weekend | 06:19 | |
| Haomeng | saripurigopi: wel | 06:21 |
| *** CHZ-PC has joined #openstack-ironic | 06:29 | |
| Haomeng | saripurigopi: another round comments:) | 06:30 |
| saripurigopi | Haomeng: sure | 06:31 |
| Haomeng | saripurigopi: dont worry, you can fix with other reviewer's comments | 06:33 |
| Haomeng | saripurigopi: £º£© | 06:34 |
| Haomeng | saripurigopi: :) | 06:34 |
| saripurigopi | Haomeng: :-) | 06:34 |
| *** Masahiro has quit IRC | 06:35 | |
| *** Masahiro has joined #openstack-ironic | 06:39 | |
| *** CHZ-PC has quit IRC | 06:43 | |
| *** saripurigopi has quit IRC | 06:47 | |
| *** rushiagr_away is now known as rushiagr | 06:49 | |
| *** openstackgerrit has quit IRC | 06:49 | |
| *** openstackgerrit has joined #openstack-ironic | 06:49 | |
| *** cuihaozhi has joined #openstack-ironic | 06:54 | |
| cuihaozhi | Hi,when i boot server from nova, nova-scheduler report "Filter RamFilter returned 0 hosts", and nova hypervisor-show $baremetal-compute display free_ram_bm=0, is there any step i missed? | 06:56 |
| cuihaozhi | util now i can use ironic node-xxx control the phy-node | 06:57 |
| rameshg87 | cuihaozhi, are you using ironic or nova-baremetal ? | 07:04 |
| cuihaozhi | i using ironic 2014.2 | 07:04 |
| cuihaozhi | with this document http://docs.openstack.org/developer/ironic/deploy/install-guide.html | 07:05 |
| cuihaozhi | if i need "ironic-nova-bm-migrate" registe some info to nova? | 07:10 |
| cuihaozhi | openstack-nova-compute.log: nova.compute.resource_tracker [-] Total physical ram (MB): 0, total allocated virtual ram (MB): 0 | 07:12 |
| rameshg87 | cuihaozhi, just wondering if you have any previous instances in nova that used this bare metal | 07:42 |
| rameshg87 | cuihaozhi, instances that have errored | 07:43 |
| rameshg87 | cuihaozhi, you could try deleting those instances which have failed with error and try | 07:43 |
| cuihaozhi | yes i delete error instance. | 07:44 |
| rameshg87 | cuihaozhi, is your ironic node having properties/memory_mb set ? | 07:45 |
| cuihaozhi | if i need ironic node-update register node meminfo? | 07:45 |
| cuihaozhi | no. | 07:45 |
| cuihaozhi | i try it/ | 07:45 |
| rameshg87 | cuihaozhi, something like this: ironic node-update $NODE_UUID add properties/cpus=4 properties/memory_mb=4096 properties/local_gb=30 properties/cpu_arch=x86_64 | 07:45 |
| rameshg87 | cuihaozhi, nova ironic virt driver reads these and sends to nova scheduler | 07:45 |
| cuihaozhi | thank u i will try it | 07:46 |
| rameshg87 | cuihaozhi, wc :) | 07:47 |
| rameshg87 | cuihaozhi, you may address a person in irc by adding their irc name to the msg (like i add your name to my msg in the beginning) | 07:48 |
| rameshg87 | cuihaozhi, because i missed your message because it didn't alert me :) | 07:48 |
| *** dtantsur|afk is now known as dtantsur | 07:59 | |
| dtantsur | Morning | 07:59 |
| openstackgerrit | Merged stackforge/ironic-discoverd: Workflow documentation is now in infra-manual https://review.openstack.org/139472 | 08:05 |
| *** jerryz has quit IRC | 08:09 | |
| *** chenglch has joined #openstack-ironic | 08:20 | |
| *** enterprisedc has joined #openstack-ironic | 08:22 | |
| *** jcoufal has joined #openstack-ironic | 08:31 | |
| *** Isotopp_ is now known as Isotopp | 08:35 | |
| *** Masahiro has quit IRC | 08:36 | |
| sirushti | dtantsur, Hi, I'd replied to your concerns and then updated the spec. Could you please take a look into https://review.openstack.org/#/c/97150/ again? | 08:37 |
| *** enterprisedc_ has joined #openstack-ironic | 08:40 | |
| dtantsur | hi, will try to (a bit later) | 08:41 |
| *** enterprisedc_ has quit IRC | 08:41 | |
| *** enterprisedc has quit IRC | 08:42 | |
| *** enterprisedc has joined #openstack-ironic | 08:44 | |
| *** Masahiro has joined #openstack-ironic | 08:47 | |
| *** dtantsur is now known as dtantsur|brb | 08:49 | |
| *** derekh has joined #openstack-ironic | 08:58 | |
| *** rakesh_hs has joined #openstack-ironic | 09:06 | |
| cuihaozhi | hi, when i boot instance.i can see ironic-conductor set "chassis bootdev pxe options=persistent", then the node just power on and not to deploy; and i can't find "dnsmasq" process on ironic-conductor, is there any thing wrong? | 09:07 |
| Haomeng | cuihaozhi: hi | 09:10 |
| Haomeng | cuihaozhi: dnsmasq is controled by neutron | 09:10 |
| Haomeng | cuihaozhi: during nova booting, ironic will call neutron update to prepare dhcp port which will launch dnsmasq process to serv the dhcp/pxe request from bm | 09:11 |
| Haomeng | cuihaozhi: so maybe your neutron issue | 09:12 |
| Haomeng | cuihaozhi: check nova ironic compute node log, and neutron logs to see what step issue during booting | 09:12 |
| cuihaozhi | Haomeng: thank u i will check it | 09:13 |
| *** jistr has joined #openstack-ironic | 09:14 | |
| *** subscope has quit IRC | 09:16 | |
| naohirot | Haomeng: Hi | 09:20 |
| naohirot | Haomeng: are you there? | 09:21 |
| *** igordcard has joined #openstack-ironic | 09:22 | |
| openstackgerrit | Tan Lin proposed openstack/ironic: Add AMT-PXE-Driver to deploy cloud on PC https://review.openstack.org/135184 | 09:23 |
| *** subscope has joined #openstack-ironic | 09:26 | |
| *** romcheg has joined #openstack-ironic | 09:27 | |
| *** andreykurilin_ has joined #openstack-ironic | 09:32 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic-specs: iLO virtual media drivers to deploy without DHCP https://review.openstack.org/137567 | 09:33 |
| *** lucasagomes has joined #openstack-ironic | 09:40 | |
| *** rakesh_hs has quit IRC | 09:45 | |
| *** foexle has joined #openstack-ironic | 09:45 | |
| *** derekh has quit IRC | 09:53 | |
| *** andreykurilin_ has quit IRC | 09:54 | |
| *** andreykurilin_ has joined #openstack-ironic | 09:54 | |
| *** subscope has quit IRC | 09:56 | |
| Haomeng | naohirot: hi | 10:03 |
| *** jerryz has joined #openstack-ironic | 10:04 | |
| *** subscope has joined #openstack-ironic | 10:06 | |
| Haomeng | naohirot: I am back:) | 10:10 |
| *** athomas has joined #openstack-ironic | 10:16 | |
| *** rakesh_hs has joined #openstack-ironic | 10:17 | |
| *** naohirot has quit IRC | 10:19 | |
| *** andreykurilin_ has quit IRC | 10:20 | |
| *** rakesh_hs2 has joined #openstack-ironic | 10:21 | |
| *** andreykurilin_ has joined #openstack-ironic | 10:21 | |
| *** rakesh_hs has quit IRC | 10:22 | |
| *** pelix has joined #openstack-ironic | 10:25 | |
| *** chenglch has quit IRC | 10:26 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic: Add tests to iscsi_deploy.build_deploy_ramdisk_options https://review.openstack.org/139097 | 10:29 |
| *** Masahiro has quit IRC | 10:33 | |
| *** rameshg87 has quit IRC | 10:39 | |
| *** rameshg87 has joined #openstack-ironic | 10:40 | |
| *** ramineni has quit IRC | 11:01 | |
| *** Haomeng has quit IRC | 11:07 | |
| *** derekh has joined #openstack-ironic | 11:09 | |
| *** Haomeng has joined #openstack-ironic | 11:10 | |
| openstackgerrit | Ramakrishnan G proposed openstack/ironic: Fix for broken deploy of iscsi_ilo driver https://review.openstack.org/139602 | 11:24 |
| *** derekh has quit IRC | 11:34 | |
| *** Masahiro has joined #openstack-ironic | 11:34 | |
| *** cuihaozhi has left #openstack-ironic | 11:36 | |
| *** Masahiro has quit IRC | 11:39 | |
| *** naohirot has joined #openstack-ironic | 11:44 | |
| *** rameshg87 has quit IRC | 11:45 | |
| *** derekh has joined #openstack-ironic | 11:47 | |
| naohirot | Haomeng: I'm back too | 11:47 |
| naohirot | Haomeng: I'm just wondering if "More than one compute_driver in nova.conf" is possible or not. | 11:48 |
| naohirot | Haomeng: http://lists.openstack.org/pipermail/openstack/2013-August/000513.html | 11:48 |
| *** lucasagomes is now known as lucas-brb | 11:51 | |
| naohirot | Haomeng: it is little bit old mail. If we configured nova for ironic, does that nova become exclusively for ironic? | 11:51 |
| openstackgerrit | Vladyslav Drok proposed openstack/ironic: Remove 'glance://' prefix strip from image hrefs https://review.openstack.org/139057 | 11:58 |
| *** derekh has quit IRC | 11:59 | |
| *** Haomeng|2 has joined #openstack-ironic | 12:01 | |
| *** Haomeng has quit IRC | 12:01 | |
| *** dtantsur|brb is now known as dtantsur | 12:06 | |
| *** andreykurilin_ has quit IRC | 12:09 | |
| *** dlaube has joined #openstack-ironic | 12:20 | |
| *** kurtrao has quit IRC | 12:22 | |
| *** kurtrao has joined #openstack-ironic | 12:22 | |
| *** rakesh_hs2 has quit IRC | 12:25 | |
| *** jcoufal has quit IRC | 12:25 | |
| *** jcoufal has joined #openstack-ironic | 12:27 | |
| *** jcoufal has quit IRC | 12:28 | |
| *** jcoufal has joined #openstack-ironic | 12:29 | |
| *** dlaube has quit IRC | 12:34 | |
| *** smoriya has quit IRC | 12:36 | |
| *** dprince has joined #openstack-ironic | 12:48 | |
| *** kurtrao has quit IRC | 12:48 | |
| *** kurtrao has joined #openstack-ironic | 12:48 | |
| Haomeng|2 | naohirot: I dont think more than one compute_driver is possible for nova compute node | 12:52 |
| Haomeng|2 | naohirot: one nova compute node support one single hypertype only | 12:52 |
| Haomeng|2 | hypervisor type | 12:52 |
| naohirot | Haomeng|2: Hi, I see | 12:53 |
| Haomeng|2 | naohirot: :) | 12:53 |
| naohirot | Haomeng|2: so typically deployer prepares the number of nova compute nodes which is same number of hyper visor deployer deals? | 12:55 |
| *** mikedillion has joined #openstack-ironic | 12:55 | |
| Haomeng|2 | naohirot: one compute node support one single hypervisor I think | 12:55 |
| Haomeng|2 | naohirot: you can confirm it from nova ironic | 12:55 |
| Haomeng|2 | naohirot: I did not test such case :) | 12:56 |
| naohirot | Haomeng|2: I see. I was asked by my colleague, but I couldn't figured out :-) | 12:56 |
| Haomeng|2 | naohirot: nova conf guide - http://docs.openstack.org/juno/config-reference/content/list-of-compute-config-options.html | 12:57 |
| Haomeng|2 | naohirot: ok | 12:57 |
| Haomeng|2 | naohirot: compute_driver = None (StrOpt) Driver to use for controlling virtualization. Options include: libvirt.LibvirtDriver, xenapi.XenAPIDriver, fake.FakeDriver, baremetal.BareMetalDriver, vmwareapi.VMwareVCDriver, hyperv.HyperVDriver | 12:59 |
| naohirot | Haomeng|2: when I looked at nova source, it seems that defining like this is impossible | 12:59 |
| *** sambetts has joined #openstack-ironic | 12:59 | |
| naohirot | Haomeng|2: compute_driver=baremetal.BareMetalDriver:libvirt.LibvirtDriver | 13:00 |
| Haomeng|2 | naohirot: sure? | 13:00 |
| Haomeng|2 | naohirot: in this doc - http://docs.openstack.org/juno/config-reference/content/list-of-compute-config-options.html | 13:00 |
| Haomeng|2 | compute_driver is not MultiStrOpt or ListOpt | 13:01 |
| Haomeng|2 | naohirot: so it should be single hypervisor type | 13:01 |
| Haomeng|2 | naohirot: can you check from nova irc room? | 13:01 |
| Haomeng|2 | good night:) I will be offline:) | 13:02 |
| Haomeng|2 | naohirot: nice weekend:) | 13:02 |
| naohirot | Haomeng|2: Okay, thanks good night:-) | 13:02 |
| naohirot | Haomeng|2: you too! | 13:02 |
| Haomeng|2 | naohirot: :) | 13:02 |
| *** Masahiro has joined #openstack-ironic | 13:08 | |
| *** bauzas has joined #openstack-ironic | 13:12 | |
| bauzas | adam_g: hi, re: https://bugs.launchpad.net/ironic/+bug/1398128 it seems that there is a regression now | 13:13 |
| bauzas | http://logs.openstack.org/73/126573/18/check/check-tempest-dsvm-ironic-pxe_ssh/a61d4c8/logs/devstacklog.txt.gz | 13:13 |
| bauzas | adam_g: http://logstash.openstack.org/#eyJzZWFyY2giOiJtZXNzYWdlOlwiTm92YSBoeXBlcnZpc29yLXN0YXRzIGRpZCBub3QgcmVnaXN0ZXIgYXQgbGVhc3QgMyBub2Rlc1wiIiwiZmllbGRzIjpbXSwib2Zmc2V0IjowLCJ0aW1lZnJhbWUiOiI0MzIwMCIsImdyYXBobW9kZSI6ImNvdW50IiwidGltZSI6eyJ1c2VyX2ludGVydmFsIjowfSwic3RhbXAiOjE0MTc3ODQzODI2NDZ9 | 13:13 |
| *** Masahiro has quit IRC | 13:13 | |
| bauzas | adam_g: I can't reopen the bug, can you ? | 13:13 |
| *** mikedillion has quit IRC | 13:19 | |
| *** pensu has quit IRC | 13:28 | |
| *** dlaube has joined #openstack-ironic | 13:34 | |
| *** pensu has joined #openstack-ironic | 13:43 | |
| *** lucas-brb is now known as lucasagomes | 13:49 | |
| openstackgerrit | Naohiro Tamura proposed openstack/ironic-specs: iRMC Management Driver for Ironic https://review.openstack.org/136020 | 13:50 |
| *** dlaube has quit IRC | 13:52 | |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: Cherry-pick changelog from 0.2.5 https://review.openstack.org/139632 | 13:56 |
| *** alexpilotti has joined #openstack-ironic | 13:56 | |
| *** jcoufal has quit IRC | 14:00 | |
| dtantsur | lucasagomes, o/ usual request to review a couple of discoverd patches: https://review.openstack.org/#/c/137374/ https://review.openstack.org/#/c/139096/ :) | 14:00 |
| *** dlaube has joined #openstack-ironic | 14:00 | |
| lucasagomes | dtantsur, hey :) will do just finishing something up here | 14:01 |
| dtantsur | ack thanks! | 14:01 |
| *** derekh has joined #openstack-ironic | 14:07 | |
| *** Marga_ has quit IRC | 14:09 | |
| *** Haomeng|2 has quit IRC | 14:11 | |
| *** pensu has quit IRC | 14:14 | |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: Do not fail if ipmi_address is not present in discovery data https://review.openstack.org/139635 | 14:14 |
| *** Haomeng|2 has joined #openstack-ironic | 14:15 | |
| *** kbyrne has quit IRC | 14:19 | |
| *** kbyrne has joined #openstack-ironic | 14:21 | |
| *** shakamunyi has joined #openstack-ironic | 14:21 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Workflow documentation is now in infra-manual https://review.openstack.org/139329 | 14:28 |
| *** romcheg has quit IRC | 14:30 | |
| *** romcheg has joined #openstack-ironic | 14:30 | |
| *** alexpilotti has quit IRC | 14:30 | |
| *** alexpilotti has joined #openstack-ironic | 14:31 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/python-ironicclient: Workflow documentation is now in infra-manual https://review.openstack.org/139374 | 14:31 |
| NobodyCam | Good morning Ironic | 14:32 |
| lucasagomes | NobodyCam, morning | 14:32 |
| dtantsur | NobodyCam, morning, looks like TGIF :) | 14:33 |
| NobodyCam | morning lucasagomes :) TGIF | 14:33 |
| openstackgerrit | Naohiro Tamura proposed openstack/ironic-specs: iRMC Virtual Media Deploy Driver for Ironic https://review.openstack.org/134865 | 14:33 |
| lucasagomes | NobodyCam, \o/ last day before holidays | 14:33 |
| NobodyCam | oh ya... morning dtantsur | 14:33 |
| NobodyCam | lucasagomes: how long are off | 14:33 |
| lucasagomes | NobodyCam, only 1 week | 14:34 |
| lucasagomes | next week I will be on holidays | 14:34 |
| lucasagomes | 8-12 of dev | 14:34 |
| lucasagomes | dec* | 14:34 |
| NobodyCam | :) oh lucasagomes is there a public picture of Pixie Boots | 14:34 |
| lucasagomes | NobodyCam, oh I should add it to the wiki right? | 14:34 |
| lucasagomes | or to our documentation? | 14:34 |
| NobodyCam | :) ya | 14:35 |
| *** romcheg has quit IRC | 14:35 | |
| lucasagomes | I will post all the inkscape files | 14:35 |
| lucasagomes | and png versions of it :) | 14:35 |
| NobodyCam | I was thinking on this page: https://wiki.openstack.org/wiki/Ironic | 14:35 |
| NobodyCam | "-p | 14:36 |
| lucasagomes | will add | 14:36 |
| jroll | morning everybody :) | 14:36 |
| NobodyCam | woo hoo :-p | 14:36 |
| NobodyCam | I've Pixie to a few people and they Love him/her/it | 14:37 |
| dtantsur | jroll, o/ | 14:37 |
| NobodyCam | mornign jroll | 14:37 |
| openstackgerrit | Merged stackforge/ironic-discoverd: Cherry-pick changelog from 0.2.5 https://review.openstack.org/139632 | 14:37 |
| *** alexpilotti has quit IRC | 14:38 | |
| naohirot | good morning, and good evening to all | 14:39 |
| NobodyCam | morning naohirot :) | 14:40 |
| *** alexpilotti has joined #openstack-ironic | 14:40 | |
| naohirot | I've updated two specs, and things to be done this week has been done :-) | 14:40 |
| lucasagomes | NobodyCam, ! | 14:40 |
| lucasagomes | jroll, morning | 14:40 |
| naohirot | NobodyCam: Hi | 14:40 |
| NobodyCam | :) | 14:41 |
| * lucasagomes is trying to edit the wiki | 14:41 | |
| dtantsur | naohirot, morning | 14:41 |
| naohirot | so I go to bed :-) | 14:41 |
| dtantsur | or is it evening? | 14:41 |
| naohirot | dtantsur: Hi :) | 14:41 |
| * dtantsur is confused | 14:41 | |
| naohirot | dtantsur: here is approaching to the mid night:-) | 14:42 |
| naohirot | so have a nice weekend to all! | 14:42 |
| *** alexpilotti has quit IRC | 14:42 | |
| dtantsur | naohirot, have a nice weekend then | 14:43 |
| naohirot | dtantsur: you too | 14:43 |
| naohirot | bye | 14:43 |
| *** naohirot has quit IRC | 14:43 | |
| NobodyCam | have a great weekend | 14:45 |
| *** mikedillion has joined #openstack-ironic | 14:45 | |
| lucasagomes | ew I got upload the pics somewhere before linking in the wiki -.- | 14:47 |
| *** ichi-the-one has joined #openstack-ironic | 14:49 | |
| ichi-the-one | HELLO nayone here? | 14:49 |
| ichi-the-one | I have a question and i will be very grateful if someone would answer me | 14:49 |
| *** erwan_taf has joined #openstack-ironic | 14:50 | |
| erwan_taf | heya world | 14:51 |
| ichi-the-one | hey erwan | 14:51 |
| openstackgerrit | Yuriy Zveryanskyy proposed openstack/ironic-specs: Add a new driver for Fuel Agent https://review.openstack.org/138115 | 14:52 |
| NobodyCam | morning ichi-the-one | 14:52 |
| NobodyCam | morning erwan_taf | 14:52 |
| ichi-the-one | hey thank you | 14:52 |
| jroll | ichi-the-one: hi :) don't ask to ask, just ask :) | 14:52 |
| ichi-the-one | cool | 14:52 |
| NobodyCam | hehhe | 14:52 |
| NobodyCam | :) | 14:52 |
| dtantsur | ichi-the-one, erwan_taf, hey! | 14:52 |
| jroll | morning(?) erwan_taf :) | 14:53 |
| ichi-the-one | i was looking into the documentation of ironic, it's mentionned that neutron is supported, is that mean that we can have isolated networks for our baremetal instances? | 14:53 |
| dlaube | g'morning guys | 14:54 |
| dtantsur | dlaube, morning | 14:54 |
| erwan_taf | lo NobodyCam, dtantsur, jroll | 14:54 |
| erwan_taf | jroll: morning yes, still in North America for a couple of hours | 14:54 |
| *** r-daneel has joined #openstack-ironic | 14:55 | |
| NobodyCam | mornign dlaube | 14:55 |
| ichi-the-one | so anyone knows the answer? | 14:55 |
| NobodyCam | ichi-the-one: all of my testing is done with a flat network | 14:57 |
| jroll | ichi-the-one: theoretically, yes. in reality, neutron does not have support today for configuring real hardware switches. | 14:57 |
| jroll | (as far as I know) | 14:57 |
| *** Masahiro has joined #openstack-ironic | 14:57 | |
| ichi-the-one | so, we have two option, use flat network with baremetal, or SDN controller to configure hardware switches as neutron can't do that, right? | 14:58 |
| jroll | ichi-the-one: sounds correct | 15:00 |
| ichi-the-one | ok, aren't there any plans to add this to neutron? | 15:00 |
| jroll | yes | 15:01 |
| * jroll finds links | 15:01 | |
| openstackgerrit | Merged openstack/ironic: Add tests to iscsi_deploy.build_deploy_ramdisk_options https://review.openstack.org/139097 | 15:01 |
| *** Masahiro has quit IRC | 15:01 | |
| jroll | ichi-the-one: maybe it's abandoned, https://review.openstack.org/#/q/topic:bp/neutron-external-attachment-points,n,z | 15:02 |
| *** alexpilotti has joined #openstack-ironic | 15:02 | |
| jroll | or moved to a different way | 15:02 |
| jroll | you should ask the neutron channel | 15:02 |
| * jroll will bbl | 15:03 | |
| ichi-the-one | ok thanks | 15:03 |
| ichi-the-one | i found this on github | 15:03 |
| ichi-the-one | https://github.com/rackerlabs/ironic-neutron-plugin | 15:03 |
| jroll | aha :) | 15:03 |
| jroll | this talks more about that plugin and how we use it https://etherpad.openstack.org/p/ironic-neutron-bonding | 15:04 |
| ichi-the-one | well thank you guys for your time | 15:04 |
| jroll | I really have to go for now but I will be back later | 15:04 |
| jroll | you're welcome :) | 15:04 |
| ichi-the-one | :) | 15:04 |
| openstackgerrit | Merged stackforge/ironic-discoverd: Do not fail if ipmi_address is not present in discovery data https://review.openstack.org/139635 | 15:05 |
| *** sambetts has quit IRC | 15:06 | |
| NobodyCam | brb | 15:12 |
| lucasagomes | dtantsur, was thinking, if we find multiple disks shouldn't we pick the first one that matches the hints? | 15:13 |
| lucasagomes | dtantsur, for example, someone just care about the disks being X size | 15:13 |
| dtantsur | lucasagomes, always a hard question. for this case probably yes. we should probably have AND of all hints, not OR (like in discoverd) | 15:14 |
| lucasagomes | dtantsur, this gives flexibility, because if someone wants to pick a very specific one he still can by passing the serial or wwn of that disk | 15:14 |
| erwan_taf | lucasagomes: _o/ | 15:14 |
| lucasagomes | erwan_taf, yo | 15:14 |
| lucasagomes | dtantsur, supporting operators? | 15:14 |
| lucasagomes | dtantsur, I was thinking about adding operators in a later version of that, talked a bit with jroll bout it | 15:15 |
| lucasagomes | like greater than, etc... we can add that stuff on top of that work | 15:15 |
| dtantsur | lucasagomes, I meant using AND by default, instead of OR | 15:15 |
| dtantsur | well, tho it still can give more than 1 variant... | 15:15 |
| erwan_taf | with mutli factor items per device be careful with and/or | 15:15 |
| lucasagomes | dtantsur, oh yeah it's AND by default :) | 15:15 |
| dtantsur | nevermind | 15:15 |
| erwan_taf | (device is a _and_ device is b) or (device_is c _and_ device is d) | 15:16 |
| erwan_taf | could make long rules | 15:16 |
| lucasagomes | erwan_taf, https://review.openstack.org/#/c/138729/ | 15:16 |
| lucasagomes | erwan_taf, yeah I'm keeping it simple for now | 15:16 |
| dlaube | so I built a custom IPA deploy image, but ran into kernel panics. then I tried using the IPA that is automatically built over at http://tarballs.openstack.org/ironic-python-agent/coreos/files/ and that is also hitting kernel panics. It there some common failure mode that I can try working around? | 15:17 |
| erwan_taf | *reading* | 15:17 |
| lucasagomes | erwan_taf, like pick the disk that matches THIS THIS THIS AND THAT | 15:17 |
| lucasagomes | no ORs | 15:17 |
| * lucasagomes brb 1 sec | 15:18 | |
| *** andreykurilin_ has joined #openstack-ironic | 15:20 | |
| *** ryanpetrello has quit IRC | 15:21 | |
| *** romcheg has joined #openstack-ironic | 15:21 | |
| *** ChuckC_ has quit IRC | 15:26 | |
| *** kurtrao has quit IRC | 15:26 | |
| *** ryanpetrello has joined #openstack-ironic | 15:26 | |
| *** kurtrao has joined #openstack-ironic | 15:26 | |
| *** Marga_ has joined #openstack-ironic | 15:27 | |
| *** romcheg has quit IRC | 15:27 | |
| *** ndipanov is now known as ndipanoff | 15:33 | |
| *** rushiagr is now known as rushiagr_away | 15:35 | |
| erwan_taf | lucasagomes: commented your proposal. Nice one, I had very few addition / comments | 15:36 |
| lucasagomes | erwan_taf, ta much! | 15:36 |
| NobodyCam | hey hey lucasagomes how familiar are you with the RH openstack packages? | 15:41 |
| lucasagomes | NobodyCam, not much :) but I can point you to the right people | 15:41 |
| lucasagomes | erwan_taf, thanks, so I was talking to dtantsur about what to do if multiple disks are found right now | 15:42 |
| NobodyCam | seems there is something strange with the nova versions.. if you have a second could you just read the replys on: https://bugs.launchpad.net/nova/+bug/1379373 | 15:42 |
| lucasagomes | erwan_taf, I was thinking that we should just pick one of the disks that matches all the expected criteria, if operator wants to pick a very specific disk he still can pass a unique hint (like uuid or wwn) | 15:43 |
| lucasagomes | NobodyCam, will do | 15:43 |
| erwan_taf | lucasagomes: +1 | 15:43 |
| NobodyCam | I'm not sure but it seems like the nova BM code is 1/2 there | 15:43 |
| lucasagomes | erwan_taf, right I will update the spec | 15:43 |
| * lucasagomes loads of comments | 15:44 | |
| lucasagomes | NobodyCam, wondering if that's specific on RH packages or if those command (nova-baremetal-manage) were actually deleted upstream as well | 15:48 |
| lucasagomes | NobodyCam, I will check internally see if I find something | 15:48 |
| erwan_taf | The best hacking scene ever : https://www.youtube.com/watch?v=boEb8zKfPBo \o/ | 15:48 |
| NobodyCam | lucasagomes: the proxy code is in place | 15:49 |
| NobodyCam | the thing I dont get is the ref to the old nova bm tables is gone. | 15:49 |
| NobodyCam | They could have a older version | 15:49 |
| lucasagomes | erwan_taf, lol ninja! | 15:49 |
| NobodyCam | but then why is the nova-baremetal-manage db sync gone? | 15:50 |
| lucasagomes | erwan_taf, I like this one https://www.youtube.com/watch?v=hkDD03yeLnU | 15:50 |
| lucasagomes | NobodyCam, yeah not sure :/ | 15:50 |
| lucasagomes | so the proxy is for Ironic right? he says he has no ironic installed | 15:51 |
| NobodyCam | yea. the command should not work | 15:52 |
| *** pcrews has joined #openstack-ironic | 15:52 | |
| NobodyCam | but its the looking for the bm tables / with out the db sync is what I'm missing | 15:53 |
| jroll | dlaube: how much ram do you have? | 15:54 |
| jroll | dlaube: you're gonna need about 1gb or more | 15:54 |
| dlaube | 16GB | 15:55 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic: Workflow documentation is now in infra-manual https://review.openstack.org/139329 | 15:55 |
| jroll | huh | 15:56 |
| jroll | dlaube: feel free to paste that somewhere, I'm about to head to my office but can look later | 15:56 |
| dlaube | ok, thanks jroll | 15:56 |
| *** killer_prince has joined #openstack-ironic | 15:57 | |
| *** killer_prince is now known as lazy_prince | 15:57 | |
| openstackgerrit | Lucas Alvares Gomes proposed openstack/ironic-specs: Root device hints https://review.openstack.org/138729 | 15:58 |
| * NobodyCam likes the cat 5 cable out of the plane | 16:03 | |
| NobodyCam | lol | 16:03 |
| * dtantsur brb | 16:06 | |
| *** rushiagr_away is now known as rushiagr | 16:07 | |
| *** ichi-the-one has quit IRC | 16:08 | |
| *** romcheg has joined #openstack-ironic | 16:10 | |
| *** killer_prince has joined #openstack-ironic | 16:13 | |
| *** lazy_prince has quit IRC | 16:14 | |
| *** killer_prince is now known as lazy_prince | 16:14 | |
| openstackgerrit | Merged stackforge/ironic-discoverd: Return serialized node to the ramdisk https://review.openstack.org/137374 | 16:17 |
| *** smoriya has joined #openstack-ironic | 16:20 | |
| *** ChuckC_ has joined #openstack-ironic | 16:24 | |
| * lucasagomes wants to use IPA and not have to deal with bash anymore :( | 16:28 | |
| *** ChuckC_ is now known as ChuckC | 16:28 | |
| * NobodyCam switchs lucasagomes shell to csh :-p | 16:29 | |
| lucasagomes | NobodyCam, it's the deploy ramdisk :( | 16:29 |
| lucasagomes | sh is it | 16:29 |
| NobodyCam | :-p | 16:29 |
| JayF | lucasagomes: TBH it probably wouldn't be terribly difficult to implement a "pxe ramdisk" in the style of IPA's (embedded container in coreos) | 16:31 |
| JayF | lucasagomes: if you'd be interested in that, I can upstream some work I did downstream to split out the build method from the build content (i.e. downstream we have >1 ramdisk so I have kinda a "ramdisk builder" library downstream) | 16:31 |
| lucasagomes | JayF, yeah it won't and we want to do that | 16:31 |
| lucasagomes | thing is, I got some priorities and can't work on that in the moment | 16:32 |
| lucasagomes | but I will! | 16:32 |
| JayF | So general question along those lines | 16:32 |
| lucasagomes | JayF, I would love to have only one ramdisk for everything | 16:32 |
| JayF | I was going to put up a spec proposing we create an ironic-ramdisks (or something similar) repository for 1) Ironic DIB elements and 2) IPA ramdisk builders not using DIB | 16:32 |
| NobodyCam | lucasagomes: +++ | 16:32 |
| JayF | lucasagomes: I'm mildly -1 to that, but primarily because it means both images will bloat larger than they need to only support one driver | 16:33 |
| lucasagomes | JayF, right, we can make it modular right? | 16:34 |
| lucasagomes | like in the DIB case we can have 1 element per module | 16:34 |
| lucasagomes | (and a base one with the core of IPA) | 16:34 |
| lucasagomes | so you can customize the ramdisk at build time | 16:34 |
| lucasagomes | and include whatever modules you want to work in ur env | 16:35 |
| JayF | I have something like that-ish setup downstream for our builder | 16:35 |
| JayF | we use image inheritance stuff from docker to make it easy though | 16:35 |
| lucasagomes | yeah so in ur case onmetal will be a module for IPA | 16:35 |
| lucasagomes | JayF, nice | 16:35 |
| lucasagomes | yeah something like that | 16:36 |
| JayF | build IPA container, from the IPA container, add "vendor" stuff from a separate dockerfile, then export the unified image as a tarball to inject into the coreos thing | 16:36 |
| JayF | yeah, I'm totally going to upstream my refactor of the builds | 16:36 |
| lucasagomes | +1 | 16:36 |
| lucasagomes | it sounds good, and very flexible | 16:36 |
| JayF | lucasagomes: would you be +1 to my idea of a separate ramdisk/elements repo for Ironic? | 16:36 |
| *** igordcard has quit IRC | 16:36 | |
| JayF | I'd love to get the IPA builder out of the repo (we build IPA out of a separate repo downstream, and it's actually simpler) and break ironic's dependence on tripleo-image-elements | 16:37 |
| lucasagomes | JayF, +1 sounds good | 16:37 |
| JayF | okay, I might find time today or this weekend to put a spec up about it | 16:38 |
| JayF | it's not a lot of work but I think would likely have a lot of benefit | 16:38 |
| *** smoriya has quit IRC | 16:38 | |
| lucasagomes | right, next week I will be a bit away (holidays) but add my name on the reviewers list so I'm aware of it once I'm back | 16:38 |
| * JayF is here except for the week of christmas | 16:39 | |
| lucasagomes | JayF, definitely. Man having to deal with bash to add something to the default ramdisk is painful :( | 16:40 |
| JayF | I really wish I could just upstream our ramdisk builder repo, as I think I have all the pieces in place to make a pxe ramdisk trivial there | 16:40 |
| JayF | but I can't because it has a crapton of NDA hardware utils in it | 16:40 |
| NobodyCam | lucasagomes: install python in the ram disk and be done with it | 16:41 |
| NobodyCam | lol | 16:41 |
| dlaube | me too JayF :) | 16:41 |
| lucasagomes | NobodyCam, can I ? :P | 16:41 |
| NobodyCam | it's prob there | 16:41 |
| lucasagomes | NobodyCam, nah I don't think so | 16:41 |
| lucasagomes | but we should heh | 16:41 |
| *** Marga_ has quit IRC | 16:43 | |
| *** igordcard has joined #openstack-ironic | 16:45 | |
| *** Masahiro has joined #openstack-ironic | 16:46 | |
| *** mjturek has quit IRC | 16:48 | |
| *** Masahiro has quit IRC | 16:50 | |
| *** yjiang5 is now known as yjiang5_away | 17:09 | |
| devananda | mornin, all | 17:10 |
| NobodyCam | good morning devananda | 17:10 |
| erwan_taf | heya devananda | 17:11 |
| erwan_taf | devananda: sounds like you made progress with the FSM idea | 17:11 |
| openstackgerrit | Jim Rollenhagen proposed openstack/ironic: Add network provider interface and implementations https://review.openstack.org/139687 | 17:14 |
| devananda | erwan_taf: to support current users, we need a migration path. to think about a migration path, we need a starting point ... | 17:14 |
| jroll | cohn: ^^ check that patch out, it's a wip but just need to shovel more code | 17:16 |
| jroll | devananda: ^ | 17:16 |
| erwan_taf | victor_lowther: I think we have a serious issue with cleaning & rebuild RAIDS in clean | 17:17 |
| erwan_taf | victor_lowther: that will prevent recognizing raids to select the proper boot device | 17:17 |
| victor_lowther | erwan_taf: how so? | 17:17 |
| victor_lowther | Nm, will reply to comment. | 17:19 |
| *** romcheg1 has joined #openstack-ironic | 17:19 | |
| *** romcheg has quit IRC | 17:20 | |
| erwan_taf | I mean if some create a RAID with the futur API, they will get something in return like UUID or similar | 17:20 |
| erwan_taf | if we do rebuild it, we're done :/ | 17:20 |
| jroll | what? | 17:21 |
| jroll | we can likely make that work | 17:21 |
| jroll | though I'm not sure what that's supposed to mean | 17:21 |
| jroll | if the node has a raid, it cleans it and rebuilds, it will store that uuid | 17:21 |
| *** andreykurilin_ has quit IRC | 17:24 | |
| *** mikedillion has quit IRC | 17:24 | |
| *** mikedillion has joined #openstack-ironic | 17:25 | |
| *** shakamunyi has quit IRC | 17:25 | |
| NobodyCam | erwan_taf: I'm not sure I see the logic error, Are you thinking of creating more then one raid volume? | 17:27 |
| devananda | jroll: why is neutron_plugin.py still ABC? just a copy-paste thing, or ? | 17:31 |
| *** romcheg has joined #openstack-ironic | 17:31 | |
| *** david-lyle has joined #openstack-ironic | 17:31 | |
| jroll | devananda: I clearly haven't implemented that yet :P | 17:31 |
| erwan_taf | NobodyCam: for sure I could need that | 17:31 |
| devananda | right | 17:31 |
| jroll | devananda: train ride ended, I pushed :P | 17:32 |
| openstackgerrit | Dmitry Tantsur proposed stackforge/ironic-discoverd: Extend node_cache.pop_node() result to be a structure https://review.openstack.org/139096 | 17:32 |
| victor_lowther | erwan_taf: for MD and DM based softwaid, you should name the array. Ditto for hardware based raid arrays that use DDF metadata. | 17:34 |
| *** foexle has quit IRC | 17:35 | |
| *** romcheg has quit IRC | 17:35 | |
| devananda | victor_lowther: just to be clear, ironic doesn't deal with software raid ... | 17:35 |
| victor_lowther | devananda: I know. | 17:36 |
| victor_lowther | Which is why the UUID question struck me as odd. | 17:37 |
| *** Marga_ has joined #openstack-ironic | 17:38 | |
| *** athomas has quit IRC | 17:39 | |
| *** Marga_ has quit IRC | 17:39 | |
| *** Marga_ has joined #openstack-ironic | 17:40 | |
| *** jmankov has quit IRC | 17:40 | |
| *** mikedillion has quit IRC | 17:40 | |
| *** jmankov has joined #openstack-ironic | 17:40 | |
| *** Marga_ has quit IRC | 17:40 | |
| *** derekh has quit IRC | 17:40 | |
| *** Marga_ has joined #openstack-ironic | 17:41 | |
| erwan_taf | victor_lowther: I mean at some point, some will ask to create RAID volumes right ? | 17:41 |
| erwan_taf | victor_lowther: and you'll have to "point it" at the bootable disk | 17:41 |
| erwan_taf | but you could also have other raids volumes and DAS disks | 17:41 |
| *** Marga_ has quit IRC | 17:41 | |
| erwan_taf | if you do delete it and recreate it in clean, how do you "point" this raid as the disk to be installed ? | 17:42 |
| *** Marga_ has joined #openstack-ironic | 17:42 | |
| *** Marga_ has quit IRC | 17:43 | |
| *** Marga_ has joined #openstack-ironic | 17:44 | |
| *** Marga_ has quit IRC | 17:44 | |
| *** jistr has quit IRC | 17:45 | |
| victor_lowther | what, you mean like megacli -adpBootDrive -set -l0 -a0 | 17:45 |
| *** Marga_ has joined #openstack-ironic | 17:45 | |
| victor_lowther | Up to whatever RAID driver you are using | 17:45 |
| *** Marga_ has quit IRC | 17:45 | |
| victor_lowther | but that is part of hte RAID config that needs to be saved and restored. | 17:45 |
| victor_lowther | or recreated | 17:46 |
| *** Marga_ has joined #openstack-ironic | 17:46 | |
| *** Marga_ has quit IRC | 17:47 | |
| NobodyCam | most raid controllers allow for a label.. | 17:47 |
| *** Marga_ has joined #openstack-ironic | 17:48 | |
| jroll | I mean | 17:48 |
| jroll | erwan_taf: we're building that into ironic right now | 17:48 |
| NobodyCam | lucasagomes: can you add disk labels to https://review.openstack.org/#/c/138729/ | 17:48 |
| jroll | hints for the disks | 17:48 |
| *** Marga_ has quit IRC | 17:49 | |
| NobodyCam | erwan_taf: see review ^^^ | 17:49 |
| jroll | and the cleaning stuff could automatically set that, if it's rebuilding a raid | 17:49 |
| *** Marga_ has joined #openstack-ironic | 17:50 | |
| openstackgerrit | Merged stackforge/ironic-discoverd: Extend node_cache.pop_node() result to be a structure https://review.openstack.org/139096 | 17:50 |
| NobodyCam | jroll: kinda like config drive partition is along the lings I was thinking | 17:50 |
| NobodyCam | lines even | 17:50 |
| jroll | yeah | 17:51 |
| lucasagomes | NobodyCam, disk label? yes | 17:51 |
| NobodyCam | :) | 17:51 |
| * lucasagomes gotta catch up with the scroll back | 17:51 | |
| devananda | so i'll preface by saying I dont have the RAID config spec paged into memory right now, but ... | 17:52 |
| *** Marga_ has quit IRC | 17:52 | |
| devananda | is that direction taking us further away from managing cattle, and closer to managing pets? | 17:53 |
| *** Marga_ has joined #openstack-ironic | 17:53 | |
| jroll | I mean, you could have a bunch of cattle with identical raid configs | 17:53 |
| devananda | victor_lowther: also, I like zehicle's of mayflies. describes the HPC use-case well | 17:53 |
| devananda | jroll: sure. in which case, that should be defined on the flavor | 17:53 |
| jroll | sure | 17:54 |
| devananda | jroll: it we put RAID config data per-node, then we're managing pets | 17:54 |
| lucasagomes | NobodyCam, apart from the label it lgty? | 17:54 |
| jroll | but you want to do it before provision time | 17:54 |
| * lucasagomes is working on it already http://paste.openstack.org/show/145628/ | 17:55 | |
| NobodyCam | lucasagomes: yea, | 17:55 |
| victor_lowther | as far as hardware RAID goes, I did a pretty deep dive on what you can see about a RAID volume using megacli | 17:55 |
| victor_lowther | (in https://github.com/opencrowbar/hardware/blob/develop/raid/crowbar_engine/barclamp_raid/app/models/barclamp_raid/lsi_megacli.rb) | 17:55 |
| *** krtaylor has quit IRC | 17:56 | |
| lucasagomes | NobodyCam, ta much, I will update it | 17:56 |
| victor_lowther | and it does not expose a UUID for RAID volumes | 17:56 |
| lucasagomes | NobodyCam, just one thing... labels have to deal with filesystem :/ | 17:57 |
| victor_lowther | much less the ability to set one. | 17:57 |
| lucasagomes | I'm not looking at that level actually | 17:57 |
| lucasagomes | why we need label ther? | 17:57 |
| lucasagomes | there* | 17:57 |
| victor_lowther | which is why I just gave the RAID volume a name | 17:57 |
| jroll | because it's useful :P | 17:57 |
| lucasagomes | jroll, right, but we are going to ovewrite it with the new image | 17:58 |
| lucasagomes | and the label will be gone | 17:58 |
| jroll | oh, true | 17:58 |
| lucasagomes | (unless the image has the same label) | 17:58 |
| lucasagomes | but still | 17:58 |
| jroll | maybe it's not useful | 17:58 |
| lucasagomes | yeah, I think it's a layer below of the one I'm using | 17:58 |
| lucasagomes | cause this is to find the device to deploy the image onto | 17:58 |
| jroll | right, you want to deal with disks, not partitions | 17:59 |
| lucasagomes | not find device for other usages | 17:59 |
| lucasagomes | yes | 17:59 |
| lucasagomes | I filter out partitons already | 17:59 |
| lucasagomes | as well as cdrom devices, ram etc | 17:59 |
| jroll | yeah, I agree, don't need labels | 17:59 |
| lucasagomes | loop devices as well | 17:59 |
| jroll | I wonder how much you're doing that IPA already does | 17:59 |
| lucasagomes | :/ | 17:59 |
| * lucasagomes too | 17:59 | |
| lucasagomes | I will do it in IPA as well | 17:59 |
| jroll | https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L226-260 | 17:59 |
| lucasagomes | but u know priorities in the moment :( | 17:59 |
| jroll | not everything, but would be easy to add walking /sys or whatever | 18:00 |
| lucasagomes | ah yeah I use lsblk for some | 18:00 |
| lucasagomes | jroll, +1 yeah that's what I'm doing | 18:00 |
| jroll | like, IPA knows about all the hardware | 18:00 |
| lucasagomes | for vendor, model size etc | 18:00 |
| lucasagomes | I look at sysfs | 18:00 |
| jroll | let's just kill the iscsi ramdisk, please | 18:00 |
| NobodyCam | lucasagomes: for HP raid cards I can :(Optional) Enter a label of no more than 15 characters to identify the array. | 18:00 |
| lucasagomes | * lucasagomes wants to use IPA and not have to deal with bash anymore :( | 18:00 |
| jroll | I can't imagine doing all of this in bash | 18:00 |
| jroll | yeah | 18:00 |
| jroll | just... do it | 18:01 |
| lucasagomes | jroll, I was talking with JayF | 18:01 |
| JayF | ipa-deploy-driver=pxe # default to agent | 18:01 |
| JayF | dooet | 18:01 |
| jroll | right | 18:01 |
| NobodyCam | that is with out a file system | 18:01 |
| jroll | but | 18:01 |
| lucasagomes | jroll, I can't, I need it asap | 18:01 |
| JayF | if the pxe driver actually uses the agent itself | 18:01 |
| JayF | that's awesome | 18:01 |
| lucasagomes | so I will have to extend a bit the deploy ramdisk first, then I can jump on IPA | 18:01 |
| jroll | JayF: that doesn't actually help you, this doesn't get hardware detection in python | 18:01 |
| * dtantsur is sick of doing discointrospecinspection in bash as well | 18:01 | |
| jroll | lucasagomes: what do you need that's not in IPA? | 18:02 |
| jroll | where you can't do this asap in IPA | 18:02 |
| *** rwsu has joined #openstack-ironic | 18:02 | |
| lucasagomes | jroll, I need the pxe_* driver to use IPA :) | 18:02 |
| jroll | right, so you need partition images and boot from pxe | 18:02 |
| lucasagomes | yup pretty much | 18:03 |
| lucasagomes | and ofc gate stuff | 18:03 |
| jroll | all we really need to do for that is rip partition code out into a library | 18:03 |
| jroll | and use it in both | 18:03 |
| lucasagomes | +1 | 18:03 |
| jroll | right, gate stuff is there, just needs approval | 18:03 |
| lucasagomes | actually, take a look at blivert | 18:03 |
| jroll | aha, yeah | 18:04 |
| lucasagomes | jroll, http://fedoraproject.org/wiki/Blivet | 18:04 |
| jroll | yeah, looking | 18:04 |
| lucasagomes | it's the anaconda stuff now in a library | 18:04 |
| lucasagomes | it seems pretty good, it can do things like creating EFI partitions and all | 18:05 |
| victor_lowther | NobodyCam: That is basically reflects what everyone else supports via the DDF | 18:05 |
| lucasagomes | JayF, if you get a time to look at it as well | 18:05 |
| *** subscope has quit IRC | 18:06 | |
| *** shakamunyi has joined #openstack-ironic | 18:07 | |
| NobodyCam | victor_lowther: I suspect thats how its doing that.. looking over the DDf stuff | 18:07 |
| lucasagomes | alright folks | 18:07 |
| lucasagomes | I will call it a day, I still have to finish my packing | 18:07 |
| lucasagomes | have a good night, a great weekend | 18:08 |
| lucasagomes | and I see y'all after next week :D | 18:08 |
| *** achanda has joined #openstack-ironic | 18:08 | |
| victor_lowther | the DDF spec is some interesting reading. | 18:08 |
| victor_lowther | I neverr knew there were so many ways to lay a RAID 5 onto the disks. | 18:08 |
| NobodyCam | have a good time off lucasagomes | 18:09 |
| lucasagomes | NobodyCam, ta much! | 18:09 |
| *** lucasagomes is now known as lucas-packing | 18:13 | |
| *** krtaylor has joined #openstack-ironic | 18:13 | |
| dtantsur | calling it a day, have a nice weekend! | 18:13 |
| dtantsur | lucas-packing, have a safe trip and enjoy your PTO | 18:13 |
| *** dtantsur is now known as dtantsur|afk | 18:13 | |
| *** pelix has quit IRC | 18:14 | |
| NobodyCam | have a good weekend dtantsur|afk | 18:14 |
| openstackgerrit | Merged openstack/ironic-specs: Workflow documentation is now in infra-manual https://review.openstack.org/139331 | 18:16 |
| *** ChuckC has quit IRC | 18:16 | |
| *** achanda has quit IRC | 18:17 | |
| *** achanda has joined #openstack-ironic | 18:18 | |
| *** pensu has joined #openstack-ironic | 18:20 | |
| *** achanda has quit IRC | 18:22 | |
| adam_g | bauzas, hey, still around? | 18:23 |
| openstackgerrit | Merged openstack/ironic: Workflow documentation is now in infra-manual https://review.openstack.org/139329 | 18:30 |
| *** harlowja_away is now known as harlowja_ | 18:31 | |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic-specs: Introduce driver capabilities https://review.openstack.org/128927 | 18:32 |
| *** jerryz has quit IRC | 18:33 | |
| *** Masahiro has joined #openstack-ironic | 18:35 | |
| *** chuckC_ has joined #openstack-ironic | 18:38 | |
| *** Masahiro has quit IRC | 18:39 | |
| *** shakamunyi has quit IRC | 18:44 | |
| *** spandhe has joined #openstack-ironic | 18:44 | |
| NobodyCam | brb | 18:46 |
| *** shakamunyi has joined #openstack-ironic | 18:46 | |
| openstackgerrit | Jay Faulkner proposed openstack/ironic-specs: Exposing Hardware Capabilities https://review.openstack.org/131272 | 18:47 |
| JayF | devananda: ^ | 18:47 |
| JayF | If others want to take a look, that's just a backlog spec, and it'd be nice to get it in ^ | 18:47 |
| *** shakamunyi has quit IRC | 18:48 | |
| devananda | +2'd | 18:48 |
| adam_g | has anyone seen conductor stall out on startup like this? http://logs.openstack.org/94/138294/7/check/check-tempest-dsvm-ironic-pxe_ssh/e3124f6/logs/screen-ir-cond.txt.gz | 18:49 |
| devananda | dtantsur's driver capabilities spec is now also backlog'd, and the two are fairly tightly related | 18:49 |
| devananda | we ought to land both of them, IMO | 18:49 |
| *** ndipanoff has quit IRC | 18:50 | |
| devananda | adam_g: that file looks truncated | 18:50 |
| *** shakamunyi has joined #openstack-ironic | 18:52 | |
| adam_g | devananda, hmm | 18:52 |
| adam_g | devananda, looking at a similiar scheduling issue as yesterday--where nova finds some enrolled nodes but they've never left power_state=None and their resources get ignored, sounded like the conductor is never syncing that on startup and thats the log i found | 18:53 |
| *** erwan_taf has quit IRC | 19:14 | |
| *** dprince_ has joined #openstack-ironic | 19:17 | |
| *** mikedillion has joined #openstack-ironic | 19:17 | |
| *** dprince has quit IRC | 19:18 | |
| *** Nisha has joined #openstack-ironic | 19:21 | |
| *** Nisha_away has joined #openstack-ironic | 19:24 | |
| *** Nisha has quit IRC | 19:26 | |
| NobodyCam | if a spec lands in backlog does that basicly mean its approved? or will it have to be revoted on to be moved in to the "current" cycle? | 19:26 |
| *** igordcard has quit IRC | 19:33 | |
| *** igordcard has joined #openstack-ironic | 19:38 | |
| NobodyCam | brb | 19:38 |
| *** pensu has quit IRC | 19:43 | |
| *** marcoemorais has joined #openstack-ironic | 19:49 | |
| *** mjturek has joined #openstack-ironic | 19:50 | |
| *** shakamunyi has quit IRC | 19:54 | |
| *** mjturek has quit IRC | 19:54 | |
| *** alexpilotti has quit IRC | 19:55 | |
| *** shakamunyi has joined #openstack-ironic | 19:59 | |
| *** shakamunyi has quit IRC | 20:00 | |
| *** shakamunyi has joined #openstack-ironic | 20:01 | |
| *** Nisha_away has quit IRC | 20:03 | |
| devananda | NobodyCam: it means there is agreement with the direction, and we want to record that. but no one's working on it now, and it is not approved for *this* cycle | 20:09 |
| devananda | NobodyCam: moving a spec out of backlog requires another round of reviews / approval | 20:09 |
| dlaube | is there a good image out there that i can build with DIB that will boot up and drop me to a root shell? | 20:10 |
| dlaube | alternatively it would be sweet if I could set pxe deploy properties to get into single user or something | 20:10 |
| dlaube | does anything like that exist? | 20:11 |
| devananda | dlaube: I believe there's a kernel param you can pass to the DIB ramdisk image which does that | 20:13 |
| devananda | or there was at one point | 20:13 |
| devananda | mm, rebooting. brb | 20:13 |
| *** dprince_ has quit IRC | 20:14 | |
| *** dprince has joined #openstack-ironic | 20:14 | |
| dlaube | hmm | 20:15 |
| *** marcoemorais has quit IRC | 20:17 | |
| NobodyCam | devananda: TY | 20:18 |
| *** Masahiro has joined #openstack-ironic | 20:23 | |
| *** Masahiro has quit IRC | 20:27 | |
| *** romcheg1 has quit IRC | 20:30 | |
| *** rushiagr is now known as rushiagr_away | 20:31 | |
| *** romcheg has joined #openstack-ironic | 20:34 | |
| jroll | dlaube: if you embed ssh keys in an IPA ramdisk, you can ssh in and do things | 20:35 |
| *** marcoemorais has joined #openstack-ironic | 20:36 | |
| dlaube | thats also a problem, the IPA i've built kernel panics and the ones I pulled off http://tarballs.openstack.org/ironic-python-agent/coreos/files/ kernel panic too | 20:37 |
| *** marcoemorais has quit IRC | 20:37 | |
| dlaube | I've set u'pxe_append_params': u'console=ttyS0' but I dont see anything on serial console | 20:37 |
| jroll | oh, right | 20:37 |
| jroll | this with a real server | 20:37 |
| jroll | ? | 20:37 |
| *** marcoemorais has joined #openstack-ironic | 20:37 | |
| dlaube | KVM over lan shows me *something* but I havent been able to catch it fly by and I dont have any backscroll | 20:38 |
| dlaube | yeah real server | 20:38 |
| jroll | you might try other params like console=ttyS1 | 20:38 |
| jroll | ah ok | 20:38 |
| jroll | hmm | 20:38 |
| *** igordcard has quit IRC | 20:38 | |
| jroll | that's interesting, I wonder if it's a coreos or kernel bug | 20:38 |
| jroll | if you can somehow get it logged, I'd love to try to help fix it, or pass it on to the coreos folks | 20:39 |
| jroll | any weird hardware or is the server pretty standard? | 20:39 |
| dlaube | pretty standard … everything is on-board nothing bleeding edge | 20:40 |
| dlaube | no HW raid or anything | 20:40 |
| jroll | huh, interesting | 20:42 |
| jroll | let me grab the latest tarball and make sure that works | 20:42 |
| jroll | (it should) | 20:42 |
| jroll | JayF: https://review.openstack.org/#/c/134436/ landed! | 20:42 |
| jroll | or +A anyway | 20:43 |
| dlaube | thanks.. I cant imagine the auto build of the image is producing a bad image but I would appeciate the check | 20:43 |
| jroll | dlaube: right, it seems to be working, see the check-tempest-dsvm-ironic-agent_ssh-nv job on https://review.openstack.org/#/c/139602/ | 20:43 |
| dlaube | does IPA work with the pxe_ipmitool driver? | 20:45 |
| jroll | no, only with agent_* | 20:45 |
| dlaube | ugghh I was afraid you would say that | 20:45 |
| jroll | I mean, that's what the agent deploy driver is for | 20:46 |
| jroll | pxe_ipmitool should really be called iscsi_ipmitool by the way | 20:46 |
| dlaube | will agent_deploy driver use ipmi ? | 20:46 |
| jroll | it can, yes | 20:46 |
| jroll | ironic in general composes deploy and power drivers (and other types) | 20:46 |
| jroll | these all use the agent deploy driver https://github.com/openstack/ironic/blob/master/setup.cfg#L38-41 | 20:47 |
| jroll | two of those have ipmi | 20:47 |
| dlaube | ahh, so it sounds like we should switch to using the agent_ipmitool | 20:47 |
| PaulCzar | so I'm real close here ... I can pxeboot guests and they run through the initial install process but then fail because they can't resolve url for my ironic api | 20:48 |
| PaulCzar | I am passing through a correct dns server via dhcp options | 20:48 |
| PaulCzar | and if I steal the MAC into another VM and boot that and get the dhcp lease that the ironic node should have ... I can ping/curl/etc the api node url | 20:48 |
| jroll | dlaube: right, the deploy driver has nothing to do with the power driver :) | 20:48 |
| NobodyCam | PaulCzar: ubuntu? | 20:48 |
| PaulCzar | NobodyCam: ya | 20:49 |
| dlaube | got it | 20:49 |
| NobodyCam | maybe somethingn with /etc/resolv.conf | 20:50 |
| PaulCzar | NobodyCam: is that something I can easily modify inside the image ? | 20:50 |
| jroll | PaulCzar: so, the ramdisk is failing, to be clear? | 20:50 |
| PaulCzar | jroll: appears so ... 'request IRONIC API to deploy image \n curl (6) Could not resolve host: openstack.example.org | 20:52 |
| dlaube | jroll: will agent_ipmitool use iscsi for deployment at all? | 20:52 |
| jroll | uhhh | 20:52 |
| jroll | PaulCzar: openstack.example.org? | 20:52 |
| jroll | dlaube: no | 20:52 |
| PaulCzar | yeah it's valid in my env | 20:52 |
| PaulCzar | our CI uses that for our dns | 20:52 |
| PaulCzar | and dnsmasq's dns server resolves it fine | 20:52 |
| jroll | dlaube: pxe_* uses iscsi, agent_* uses IPA and thus http/dd | 20:52 |
| jroll | PaulCzar: ok, yeah | 20:52 |
| jroll | seems like a diskimage-builder thing, seems awful weird | 20:53 |
| dlaube | jroll: ok thanks for clarifying | 20:53 |
| jroll | I'd probably bug #tripleo | 20:53 |
| jroll | dlaube: np | 20:53 |
| PaulCzar | jroll: somebody mentioned the other day that there's a cirros test image? is that just the standard image? if so what do I use for vmlinuz/ramdisk/etc ? | 20:57 |
| jroll | aweeks: you should look at https://review.openstack.org/#/q/status:open+project:openstack/ironic+branch:master+topic:states,n,z | 20:57 |
| jroll | PaulCzar: it can be broken up into kernel/ramdisk, sec | 20:57 |
| aweeks | jroll: looking | 20:57 |
| *** shakamunyi has quit IRC | 20:59 | |
| jroll | PaulCzar: https://launchpad.net/cirros/+download | 21:00 |
| jroll | the -uec images are split | 21:00 |
| dlaube | jroll: in order to use http + dd to deploy via the agent_ipmi driver, I dont need to use iPXE do I? | 21:03 |
| *** mikedillion has quit IRC | 21:03 | |
| jroll | dlaube: nope, agent does what it does regardless of how you boot it :) | 21:04 |
| jroll | actually, I don't know if there's ipxe support for the agent driver yet (which is a bit silly) | 21:04 |
| dlaube | ok cool | 21:05 |
| dlaube | ty! | 21:05 |
| *** romcheg has quit IRC | 21:05 | |
| jroll | np :) | 21:05 |
| *** shakamunyi has joined #openstack-ironic | 21:06 | |
| JayF | agent_ssh-src is voting on IPA \o/ https://review.openstack.org/#/c/134436/ | 21:06 |
| *** spandhe has quit IRC | 21:06 | |
| *** dprince has quit IRC | 21:08 | |
| *** ChuckC has joined #openstack-ironic | 21:10 | |
| PaulCzar | jroll: do I need to untar the uec and add them individually ... or will ironic know how to handle the uec image ? | 21:12 |
| jroll | PaulCzar: I think untar and add manually | 21:13 |
| PaulCzar | k, and are the kernel/ramdisk image IDs added to the glance image metadata different to the ones set in the ironic node options ( -i pxe_deploy_kernel= ) ? | 21:15 |
| PaulCzar | or can I use them for both ? | 21:15 |
| jroll | I have no idea, I've never done this before | 21:17 |
| jroll | I'm not entirely sure what you're asking | 21:17 |
| jroll | oh | 21:17 |
| PaulCzar | haha nobody has it seems :D | 21:17 |
| jroll | so, upload the cirros k/r to glance | 21:17 |
| jroll | use those ids as the metadata for the cirros root image | 21:17 |
| *** marcoemorais has quit IRC | 21:17 | |
| jroll | the root image points at its own kernel and ramdisk, hopefully that makes sense | 21:17 |
| *** marcoemorais has joined #openstack-ironic | 21:18 | |
| PaulCzar | okay, so I shouldn't need the -i pxe_deploy_kernel= -i pxe_deploy_ramdisk options in my ironic node-create command then ? | 21:18 |
| jroll | correct, ironic handles that | 21:19 |
| PaulCzar | right | 21:19 |
| NobodyCam | brb quick walkies | 21:23 |
| openstackgerrit | Merged openstack/ironic-specs: Exposing Hardware Capabilities https://review.openstack.org/131272 | 21:23 |
| *** romcheg has joined #openstack-ironic | 21:25 | |
| bauzas | adam_g: hey I'm there | 21:29 |
| *** ChuckC has quit IRC | 21:30 | |
| *** chuckC_ has quit IRC | 21:31 | |
| aweeks | jroll: devananda: so, from my reading of the three FSM related changes out right now, we're not actually using the FSM to control how state transitions are handled. As in, the FSM is being used to verify that a state transition is valid, but not to actually trigger the action as a result of the state transition. | 21:31 |
| devananda | aweeks: baby steps .... :) | 21:32 |
| aweeks | heh, ko | 21:32 |
| aweeks | *ok | 21:32 |
| devananda | aweeks: we aren't using *any* state machine today. but I think those three lay enough down that I can start using task_manager / fsm.py to do callbacks | 21:32 |
| devananda | thus the refactoring of eg, _do_node_deploy and so on | 21:33 |
| aweeks | yeah, that makes sense | 21:33 |
| aweeks | awesome | 21:33 |
| devananda | if you feel like having a go at that, lemme know | 21:33 |
| aweeks | devananda: this might be more of a philosophical question, but do you think that the conductor *is* a state machine (or really multiple state machines, one for each node), or that it simply is *implemented* as a state machine. not sure if that makes sense | 21:35 |
| *** dlaube has quit IRC | 21:36 | |
| aweeks | mostly having to do with the api--is the contract that we are a state machine, and you tell us how to transition? | 21:36 |
| devananda | oh | 21:36 |
| aweeks | or is the relationship between api calls that the state machine looser than that | 21:36 |
| devananda | I'll rephrase - see if this is what you meant | 21:36 |
| adam_g | bauzas, hi, regarding that scheduling issue you saw. is that the first time you've hit that? ive been looking into some other similar weirdness nova's resource tracking with ironic | 21:37 |
| devananda | "ConductorManager is the state machine; it accepts input, determines what state to transition to, and then transitions to it" | 21:37 |
| devananda | vs | 21:37 |
| bauzas | adam_g: well, I just discovered the Ironic problem because of the Nova job which was failing on my patch | 21:38 |
| devananda | "ConductorManager tracks the state of each node in a logical state machine, but transitions are handled externally (eg, by the drivers)" | 21:38 |
| devananda | aweeks: not sure my words are any clearer :( | 21:39 |
| bauzas | adam_g: indeed, the Nova RT is by design providing updates to the sched every 60 secs | 21:39 |
| aweeks | devananda: no, I think we're on the same page | 21:39 |
| aweeks | I'm just curious which vision you have have right now | 21:39 |
| bauzas | adam_g: so, the question I'm wondering is why it takes more 120 secs for deploying 3 ironic nodes | 21:39 |
| adam_g | bauzas, yeah, i looked into that specific failure case you saw and it looks like an issue on the ironic side (nova's RT is discovering nodes in the ironic inventory, but ironic's conductor never syncs them to a usable state) | 21:40 |
| adam_g | bauzas, one sec, let me pull up those logs again | 21:40 |
| devananda | aweeks: #2 | 21:40 |
| bauzas | adam_g: oh ok | 21:40 |
| aweeks | kk | 21:40 |
| devananda | aweeks: logical state machine. Conductor instantiates one state machine for each node it manages, uses that object to validate events and initiate(*) transitions | 21:41 |
| devananda | aweeks: (*) I need to add callbacks to actually do this | 21:41 |
| aweeks | devananda: I guess another way to think about this is whether the API is more state oriented, or more transition oriented. | 21:41 |
| devananda | aweeks: which api, to be precise? | 21:42 |
| aweeks | my understanding is there are two apis involved: | 21:42 |
| aweeks | 1. you would set the provision state in ironic api | 21:43 |
| aweeks | 2. it would make an rpc call to the conductor | 21:43 |
| aweeks | the first one is more state oriented (you tell it what state you want it in), the second one is more transition orinted (you would call do_deploy, or do_spawn, etc.) | 21:43 |
| devananda | aweeks: hm. I think you've drawn that line in the wrong place | 21:44 |
| devananda | aweeks: the RPC call is essentially just routing the API request to the appropriate conductor. it's not really different, though | 21:45 |
| aweeks | 1 sec, let me double check the code | 21:45 |
| devananda | aweeks: within the ConductorManager.do_node_deploy(), for example, there are multiple calls out to the driver, to utils, etc | 21:45 |
| devananda | aweeks: each of those calls performs a transition. some are micro-state (eg, DEPLOYING -> DEPLOYDONE) and others are macro-state (AVAILABLE -> DEPLOYING) | 21:46 |
| devananda | right now, the Conductor handles updating the node.*_state fields based on a) requests to start an action, and b) return values from other libraries (drivers, utils, etc) | 21:47 |
| devananda | so, in that sense, yes, the ConductorManager today *IS* the state machine | 21:47 |
| adam_g | bauzas, so it shouldn't take more than 120 secs for nodes to register. it typically takes less, 2 minutes was generous to give a grenade nova-bm -> ironic time to migrate nodes into ironic. something went wrong on the ironic side and resources never ended up in nova | 21:47 |
| devananda | but as such, not clearly defined. within the API service, there is some state checking (and it can get out-of-sync w.r.t. the conductor's state checking ....) | 21:48 |
| aweeks | got it | 21:48 |
| devananda | aweeks: and the state transitions are implicit within the conductor and driver code, not formally modelled in a class somewhere --- that is the biggest change my patch series does | 21:48 |
| adam_g | bauzas, nova will check ironic for resources at startup, and then update during periodic tasks. the devstack polling for in the case that nodes aren't enrolled until after the initial startup resource update, and we need to wait for a subsequent periodic task | 21:49 |
| devananda | though it's not complete yet. there are still places making implicit state checks (eg, the API service) which aren't using fsm.machine | 21:49 |
| devananda | ** states.machine, i mean | 21:49 |
| bauzas | adam_g: k, see the problem | 21:49 |
| adam_g | bauzas, feel free to recheck, i'd like to get a bug filed and an e-r query in case this turns up agani | 21:50 |
| bauzas | adam_g: I guess we can probably update the existing e-r query | 21:51 |
| adam_g | bauzas, this is a different bug | 21:52 |
| adam_g | bauzas, incidentally, the existing bug is still showing up but for other unknown RT reasons | 21:52 |
| bauzas | adam_g: what are your symptoms ? | 21:53 |
| bauzas | adam_g: I'm working on the RT/Scheduler interfaces in Nova so I can probably help | 21:53 |
| aweeks | devananda: so, from my possibly uninformed perspective, I've been thinking about nodes in ironic as being controlled by state machine, of which the nova state machine is a subset. we have our own set of states, along with a view of state that is shown to nova. | 21:53 |
| devananda | aweeks: right. that's actually inverted | 21:54 |
| adam_g | bauzas, cool! let me update Bug #1398128 with the details | 21:54 |
| devananda | aweeks: logically speaking, set{ironic_states} intersects set{nova_states} | 21:55 |
| devananda | aweeks: there are states that Nova can represent to its users which do not apply to Ironic (eg, migrating, snapshotting) | 21:55 |
| bauzas | adam_g: k, will look at logstash for the frequency | 21:55 |
| devananda | aweeks: and there are states within Ironic which Nova does not model in any way (eg, discovering, cleaning, managed, or available) | 21:56 |
| *** lucas-packing has quit IRC | 21:56 | |
| devananda | aweeks: functionally, each service is modelling a different type of resource. Further, Ironic actually has no direct knowledge of what state the corresponding Nova instance is in | 21:57 |
| aweeks | devananda: ok, that makes sense | 21:57 |
| aweeks | all we get is the nova calls through the virt driver | 21:57 |
| devananda | yup | 21:57 |
| devananda | which implement a very small set of primitives | 21:58 |
| aweeks | which might implicitly tell us something | 21:58 |
| aweeks | cooincidentally | 21:58 |
| devananda | yup | 21:58 |
| devananda | eg, nova.virt.driver.spawn() implies that there is a Nova instance | 21:58 |
| devananda | and we should save certain information in Ironic to indicate the association of that instance with that node | 21:59 |
| aweeks | yep | 21:59 |
| devananda | so that Nova can find it again later | 21:59 |
| *** marcoemorais has quit IRC | 22:01 | |
| *** marcoemorais has joined #openstack-ironic | 22:01 | |
| *** marcoemorais has quit IRC | 22:01 | |
| *** marcoemorais has joined #openstack-ironic | 22:01 | |
| *** marcoemorais has quit IRC | 22:02 | |
| *** marcoemorais has joined #openstack-ironic | 22:02 | |
| *** ryanpetrello_ has joined #openstack-ironic | 22:06 | |
| *** ryanpetrello has quit IRC | 22:06 | |
| *** ryanpetrello_ is now known as ryanpetrello | 22:06 | |
| *** igordcard has joined #openstack-ironic | 22:09 | |
| adam_g | bauzas, okay, added some notes to https://bugs.launchpad.net/ironic/+bug/1398128 found the race as i was looking at the logs in more detail | 22:11 |
| *** Masahiro has joined #openstack-ironic | 22:12 | |
| *** ryanpetrello has quit IRC | 22:14 | |
| bauzas | adam_g: that's really weird | 22:15 |
| bauzas | adam_g: that's the same section of code in the RT | 22:15 |
| bauzas | adam_g: there is only a branch if the compute is present or not | 22:16 |
| *** Masahiro has quit IRC | 22:17 | |
| bauzas | adam_g: do you know how the Ironic driver reports stats to Nova ? | 22:17 |
| bauzas | adam_g: I basically know that 1 host can run multiple nodes | 22:17 |
| bauzas | which is not the same for other drivres | 22:18 |
| adam_g | bauzas, yeah, so it makes an api call to ironic to list nodes. each node is a nova hypervisor. each ironic node has properties associated with it describing mem/cpu/gb, which the driver maps to nova's resources | 22:18 |
| adam_g | bauzas, it looks like, when nova first finds some nodes in ironic, it sets the hypervisor count accordingly but doesn't update associated resources until the next periodic sync | 22:19 |
| adam_g | at least according to the symptoms, looking thru nova code now | 22:19 |
| *** spandhe has joined #openstack-ironic | 22:24 | |
| adam_g | bauzas, whats confusing me is how we're ending up with 0 resources here: http://logs.openstack.org/87/139687/1/check/check-tempest-dsvm-ironic-parallel-nv/55ce684/logs/screen-n-cpu.txt.gz#_2014-12-05_18_01_26_852 | 22:27 |
| bauzas | adam_g: I should look at the driver's code in Nova... | 22:29 |
| bauzas | adam_g: that's 11.30pm here, could we catch up on Monday morning your time ? | 22:30 |
| adam_g | bauzas, sure! still early here, check back at that bug i'll update it with anything i find this afternoon | 22:30 |
| bauzas | adam_g: which TZ are you in ? | 22:30 |
| adam_g | bauzas, PST USA | 22:30 |
| bauzas | adam_g: I'm CET | 22:30 |
| bauzas | adam_g: ie. France TZ | 22:31 |
| bauzas | adam_g: ok, so I'll try to find some time to investigate on Monday | 22:31 |
| adam_g | bauzas, cool, have a good weekend | 22:31 |
| NobodyCam | oh thats new: error: Failed to start domain seed | 22:38 |
| NobodyCam | error: Unable to add port vnet1 to OVS bridge brbm: Operation not permitted | 22:38 |
| *** alexpilotti has joined #openstack-ironic | 22:49 | |
| *** r-daneel has quit IRC | 22:49 | |
| PaulCzar | if I set the ironic-api to IP then I can actually boot an instance! but it takes forever because it can't reach metadata .. is there a SNAT or similar trick to allow access to metadata on the 169. address ? | 22:50 |
| *** andreykurilin_ has joined #openstack-ironic | 22:52 | |
| NobodyCam | PaulCzar: TripleO does this: https://github.com/openstack/tripleo-image-elements/blob/master/elements/nova-ironic/os-refresh-config/configure.d/81-nat-metadata#L6 | 22:56 |
| *** ChuckC has joined #openstack-ironic | 22:59 | |
| *** alexpilotti has quit IRC | 23:00 | |
| *** shakamunyi has quit IRC | 23:01 | |
| *** Haomeng has joined #openstack-ironic | 23:06 | |
| *** Haomeng|2 has quit IRC | 23:07 | |
| PaulCzar | NobodyCam: is that on the guest ? | 23:13 |
| NobodyCam | PaulCzar: yep: os-refresh-config is baked into the image and run after deploy | 23:14 |
| NobodyCam | SpamapS: are there better words for os-refresh-config (see ^) | 23:15 |
| *** alexpilotti has joined #openstack-ironic | 23:20 | |
| *** spandhe has left #openstack-ironic | 23:23 | |
| SpamapS | NobodyCam: yeah, os-refresh-config is run by os-collect-config .. but that is irrelevant really. | 23:23 |
| NobodyCam | :) | 23:23 |
| *** rwsu has quit IRC | 23:24 | |
| NobodyCam | Ty SpamapS | 23:25 |
| SpamapS | NobodyCam: btw, that is not on the guest. | 23:27 |
| SpamapS | NobodyCam: that is on the conductor | 23:27 |
| * NobodyCam has a headach after spending way to much time trouble shooting swift post -m ‘Temp-URL-Key:unset’ vs swift post -m "Temp-URL-Key:unset" | 23:27 | |
| NobodyCam | oh good catch | 23:27 |
| SpamapS | PaulCzar: ^^ | 23:27 |
| NobodyCam | PaulCzar: see SpamapS comment ^^ | 23:28 |
| *** andreykurilin_ has quit IRC | 23:28 | |
| SpamapS | PaulCzar: your instance should be getting a pushed route via DHCP | 23:30 |
| SpamapS | PaulCzar: we set a DHCP option in Neutron to tell your instance that 169.254.169.254 routes through $NETWORK_NODE | 23:30 |
| NobodyCam | SpamapS: i don't think PaulCzar is running TripleO scripts | 23:31 |
| SpamapS | PaulCzar: the NAT rule above just translates that to the local neutron-metadata-agent | 23:31 |
| SpamapS | NobodyCam: I thought that was what the ironic nova driver did when making ports. | 23:32 |
| SpamapS | I could be wrong, might be a config option or something. | 23:32 |
| NobodyCam | i don't think we push routes | 23:32 |
| NobodyCam | thou imbw | 23:32 |
| NobodyCam | too | 23:32 |
| NobodyCam | oh SpamapS side question you may just know the answer to | 23:33 |
| *** marcoemorais has quit IRC | 23:33 | |
| NobodyCam | should os-refresh-config/migration.d scripts retry if they exit with other then 1 | 23:33 |
| *** marcoemorais has joined #openstack-ironic | 23:34 | |
| NobodyCam | *other then 0 | 23:34 |
| SpamapS | NobodyCam: o-r-c just explodes when that happens | 23:35 |
| SpamapS | NobodyCam: but os-collect-config will indeed retry over and over again. | 23:35 |
| NobodyCam | heheh think I found a bug. then lol | 23:35 |
| NobodyCam | occ reruns orc but then something in os-config-refresh/post-configure.d/99-refresh-completed fails | 23:37 |
| NobodyCam | don't waste any thought on it.. it was a bad approch... | 23:38 |
| *** igordcard has quit IRC | 23:42 | |
| SpamapS | NobodyCam: That's intentional that it keeps running over and over until everything succeeds with the latest version of metadata. | 23:43 |
| SpamapS | NobodyCam: o-r-c scripts _MUST_ be idempotent. | 23:43 |
| * jroll working on https://bugs.launchpad.net/nova/+bug/1399830 | 23:48 | |
| NobodyCam | SpamapS: see http://paste.openstack.org/show/RD3Wi5Iyu7PsNDMKte6z/ | 23:48 |
| NobodyCam | jroll: awesome TY | 23:49 |
| jroll | NobodyCam: pretty nasty bug :/ | 23:49 |
| NobodyCam | :( | 23:49 |
| NobodyCam | SpamapS: as a result of what that paste shows the migration.d script never runs again | 23:50 |
| SpamapS | NobodyCam: that's unfortunate, but it is part of the fact that the signal failed.. which is a serious issue. | 23:52 |
| NobodyCam | :) it was a bad approch | 23:53 |
| adam_g | if conductor is is blocking on 'Attempting to reserve node 5', is there an easy way to find out what task has that node locked? | 23:55 |
| NobodyCam | grep the log for the nodes uuid | 23:56 |
| adam_g | NobodyCam, not much to grep for, http://logs.openstack.org/94/138294/7/check/check-tempest-dsvm-ironic-pxe_ssh/e3124f6/logs/screen-ir-cond.txt.gz | 23:57 |
| adam_g | a different log than what im working with, but the same issue | 23:57 |
| devananda | adam_g: what's the issue? | 23:58 |
| devananda | adam_g: I don't see any error | 23:58 |
| adam_g | devananda, its the that log we saw yesterday ^^, seems to be related to the resource tracking/scheduling issues that are popping up | 23:58 |
| adam_g | nova's node.list/get returns all of the nodes, but none of them have a power state synced because conductor is stuck waiting for a reservation | 23:59 |
| adam_g | without a power state, nova skips them for scheduling | 23:59 |
| devananda | going through the other logs in that run ... this looks interesting | 23:59 |
| devananda | 2014-12-05 10:11:38.819 31500 INFO urllib3.connectionpool [-] Resetting dropped connection: 127.0.0.1 | 23:59 |
| devananda | in ir-api | 23:59 |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!
