| *** bdodd has quit IRC | 00:00 | |
| *** rcernin has joined #openstack-ironic | 00:05 | |
| *** rcernin has quit IRC | 00:06 | |
| *** bdodd has joined #openstack-ironic | 00:12 | |
| stevebaker | good afternoon | 00:41 |
|---|---|---|
| openstackgerrit | Christopher Dearborn proposed openstack/ironic master: Redfish driver firmware update https://review.opendev.org/749619 | 00:48 |
| *** rloo has quit IRC | 01:24 | |
| *** tzumainn has quit IRC | 01:26 | |
| *** ianychoi_ has joined #openstack-ironic | 01:46 | |
| *** ianychoi has quit IRC | 01:48 | |
| *** iurygregory has quit IRC | 02:08 | |
| *** rh-jelabarre has quit IRC | 03:12 | |
| *** rh-jelabarre has joined #openstack-ironic | 03:13 | |
| *** cdearborn has quit IRC | 03:17 | |
| *** rh-jelabarre has quit IRC | 03:18 | |
| *** Lucas_Gray has joined #openstack-ironic | 03:31 | |
| *** JohnMice has joined #openstack-ironic | 03:45 | |
| *** mkrai has joined #openstack-ironic | 03:52 | |
| *** Qianbiao has joined #openstack-ironic | 04:00 | |
| *** JamesBenson has quit IRC | 04:09 | |
| *** JamesBenson has joined #openstack-ironic | 04:11 | |
| *** JamesBenson has quit IRC | 04:50 | |
| *** uzumaki has joined #openstack-ironic | 05:06 | |
| uzumaki | good morning ironic! o/ | 05:08 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Utility functions for REST API JSON handling https://review.opendev.org/751160 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert allocations endpoint to plain JSON https://review.opendev.org/743679 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert bios endpoint to plain JSON https://review.opendev.org/745475 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert chassis endpoint to plain JSON https://review.opendev.org/745476 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert conductors endpoint to plain JSON https://review.opendev.org/745779 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert deploy_templates endpoint to plain JSON https://review.opendev.org/745780 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert drivers endpoint to plain JSON https://review.opendev.org/741865 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert event endpoint to plain JSON https://review.opendev.org/746623 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert nodes endpoint to plain JSON https://review.opendev.org/747820 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert ports endpoint to plain JSON https://review.opendev.org/750120 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert portgroups endpoint to plain JSON https://review.opendev.org/750864 | 05:19 |
| *** zzzeek has quit IRC | 05:19 | |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert ramdisk endpoint to plain JSON https://review.opendev.org/750857 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert volume endpoint to plain JSON https://review.opendev.org/750858 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert volume/connectors endpoint to plain JSON https://review.opendev.org/750859 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Convert volume/targets endpoint to plain JSON https://review.opendev.org/750860 | 05:19 |
| openstackgerrit | Steve Baker proposed openstack/ironic master: Expunge the internal version of WSME https://review.opendev.org/750861 | 05:20 |
| *** abdysn has joined #openstack-ironic | 05:21 | |
| *** zzzeek has joined #openstack-ironic | 05:22 | |
| *** JamesBenson has joined #openstack-ironic | 05:31 | |
| *** JamesBenson has quit IRC | 05:36 | |
| *** Qianbiao has quit IRC | 05:42 | |
| *** stevebaker has quit IRC | 05:45 | |
| *** stevebaker has joined #openstack-ironic | 05:46 | |
| *** jtomasek has joined #openstack-ironic | 06:03 | |
| *** dking has quit IRC | 06:06 | |
| *** dking has joined #openstack-ironic | 06:09 | |
| *** Lucas_Gray has quit IRC | 06:15 | |
| *** Qianbiao has joined #openstack-ironic | 06:27 | |
| arne_wiebalck | Good morning, ironic! | 06:33 |
| *** dking has quit IRC | 06:42 | |
| *** zzzeek has quit IRC | 06:45 | |
| *** zzzeek has joined #openstack-ironic | 06:47 | |
| Qianbiao | Good morning ironic. | 06:52 |
| Qianbiao | Morning arne_wiebalck | 06:52 |
| arne_wiebalck | Good morning, Qianbiao o/ | 06:53 |
| Qianbiao | hey o/ you are earlier today. | 06:53 |
| *** mkrai has quit IRC | 06:53 | |
| Qianbiao | Do u have a minute look at this story： https://storyboard.openstack.org/#!/story/2008142, am I correct? | 06:53 |
| * arne_wiebalck is reading ... | 06:56 | |
| *** eandersson has quit IRC | 06:58 | |
| *** iurygregory has joined #openstack-ironic | 06:58 | |
| iurygregory | good morning Ironic! | 06:59 |
| *** dking has joined #openstack-ironic | 06:59 | |
| *** eandersson has joined #openstack-ironic | 06:59 | |
| arne_wiebalck | From what I understand, hardware managers are not invoked based on the found hardware. They are detected/found, and when functions need to be called, they are called in the order of the priority of the hardware managers. | 07:00 |
| *** zzzeek has quit IRC | 07:01 | |
| *** eandersson has quit IRC | 07:01 | |
| arne_wiebalck | So, if your IPA has a special hardware manager built in, it will check it first for any function to be called before it goes to the generic h/w manager. | 07:01 |
| Qianbiao | arne_wiebalck it seems the generic hardware is run before mlnx hardware. | 07:02 |
| *** eandersson has joined #openstack-ironic | 07:02 | |
| arne_wiebalck | This can happen if a function is not overwritten in the mlnx one. | 07:02 |
| arne_wiebalck | But again, this is how I understand things work. | 07:03 |
| *** zzzeek has joined #openstack-ironic | 07:03 | |
| arne_wiebalck | We have a downstream hw manager and this is how I used it. | 07:03 |
| Qianbiao | arne_wiebalck look at this: https://github.com/openstack/ironic-python-agent/blob/master/ironic_python_agent/hardware.py#L2095 | 07:04 |
| Qianbiao | the hardware call order is restrict to this compare functino. | 07:04 |
| Qianbiao | is it wrong? | 07:05 |
| arne_wiebalck | No, that is correct. But imagine you have call a function foo() which is not in the higher prio manager. It will then be called in the lower priority one. | 07:06 |
| arne_wiebalck | You see what I mean. | 07:06 |
| arne_wiebalck | ? | 07:06 |
| arne_wiebalck | The managers are ordered, but if some of them do not implement a function, we fall back to the lower prio one. This may look like they are called out of order. | 07:07 |
| arne_wiebalck | But they are not. | 07:07 |
| Qianbiao | yes, you are right. | 07:08 |
| arne_wiebalck | You may want to switch on debugging to see which functions are called in which order. | 07:09 |
| arne_wiebalck | https://docs.openstack.org/ironic-python-agent/latest/contributor/hardware_managers.html | 07:10 |
| arne_wiebalck | section "How are methods executed on HardwareManagers?" | 07:10 |
| arne_wiebalck | The other thing to check is whether your IPA really uses the mlnx hardware manager. | 07:11 |
| arne_wiebalck | Or only the generic one. | 07:11 |
| arne_wiebalck | Not sure if manager leave a log message automatically ... our downstream one does, so we know it is executed. I can check where we do this ... | 07:12 |
| Qianbiao | <arne_wiebalck> a big problem to me now is: inspect ramdisk log is truncated, | 07:13 |
| Qianbiao | i do not know why, everytime it only send a small part of log back. | 07:13 |
| Qianbiao | Morning iurygregory | 07:14 |
| arne_wiebalck | Qianbiao: in a meeting now, will get back to you after | 07:14 |
| Qianbiao | thanks !! | 07:14 |
| iurygregory | hey Qianbiao o/ | 07:20 |
| Qianbiao | Hey o/ | 07:21 |
| Qianbiao | arne_wiebalck I find the issue, the first run hardware is IntelCnaHardwareManager, and it implements GenericHardwareManager, then mlnx hardware_mgr will never run, if hw supports IntelCnaHardwareManager | 07:22 |
| *** ricolin has quit IRC | 07:32 | |
| Qianbiao | I update the story, it maybe more clear now. | 07:32 |
| *** JamesBenson has joined #openstack-ironic | 07:32 | |
| *** tosky has joined #openstack-ironic | 07:34 | |
| *** JamesBenson has quit IRC | 07:36 | |
| *** priteau has joined #openstack-ironic | 07:39 | |
| *** ricolin has joined #openstack-ironic | 07:42 | |
| openstackgerrit | Verification of a change to openstack/ironic failed: Add release note for dhcp-less deploy https://review.opendev.org/751139 | 07:49 |
| Qianbiao | hello iurygregory, if i want to update an old branch. | 07:53 |
| openstackgerrit | zhufl proposed openstack/ironic master: Fix invalid assert_has_calls https://review.opendev.org/750910 | 07:54 |
| Qianbiao | I just git checkout stable/train; git checkout -b my-topic. | 07:54 |
| Qianbiao | then update code and git review? | 07:54 |
| Qianbiao | it will auto match to the stable branch? | 07:54 |
| iurygregory | Qianbiao, if the change only needs to go to stable/branch yes | 07:54 |
| Qianbiao | ok thanks. | 07:55 |
| iurygregory | if is something you want on master first you send the patch to master, than you do the backport | 07:55 |
| Qianbiao | sure, got. | 07:56 |
| *** Abdallahyas has joined #openstack-ironic | 07:58 | |
| *** Abdallahyas has quit IRC | 08:00 | |
| *** abdysn has quit IRC | 08:01 | |
| *** mkrai has joined #openstack-ironic | 08:02 | |
| *** sshnaidm|afk is now known as sshnaidm | 08:04 | |
| openstackgerrit | Verification of a change to openstack/ironic failed: Remove install unnecessary packages https://review.opendev.org/751636 | 08:11 |
| *** lucasagomes has joined #openstack-ironic | 08:13 | |
| arne_wiebalck | Qianbiao: I guess you can have more than 2 managers even. | 08:22 |
| Qianbiao | arne_wiebalck not understand. | 08:23 |
| Qianbiao | I think it's clear that there is a bug in ipa, right? | 08:24 |
| arne_wiebalck | Qianbiao: the managers are ordered by priority, so even if you have the intel one it does not mean you cannot have the mlnx one. | 08:24 |
| Qianbiao | <arne_wiebalck> the intel one will do everything. | 08:25 |
| Qianbiao | then, the mlnx one will never run. | 08:25 |
| arne_wiebalck | correct | 08:25 |
| Qianbiao | because it has all interface method. | 08:26 |
| arne_wiebalck | yes | 08:26 |
| arne_wiebalck | that is my understanding as well | 08:26 |
| Qianbiao | what is your suggestion to fix this. | 08:26 |
| Qianbiao | I am not quite understand "I guess you can have more than 2 managers even." | 08:26 |
| arne_wiebalck | what do you have the intel hw manager for? | 08:26 |
| arne_wiebalck | I mean, why do you use it? | 08:27 |
| Qianbiao | not sure, our stack hardware all have it. | 08:27 |
| arne_wiebalck | Right. So, my understanding is that a hw manager is supposed to manage specific hardware and it seems the intel one is not the correct one for the nodes you're trying to manage. | 08:28 |
| Qianbiao | I think it's bug, the intel one should not inherit generic hw mgr. | 08:29 |
| Qianbiao | if it only different in "collect_lldp_data", it should only implement "collect_lldp_data" function. | 08:30 |
| arne_wiebalck | I think all hardware managers need to implement evaluate_hardware_support(), no? | 08:31 |
| arne_wiebalck | Yes. | 08:31 |
| Qianbiao | yes | 08:31 |
| Qianbiao | it must | 08:31 |
| arne_wiebalck | What ours does is to also call the evaluate_hardware_support() of the parent class. | 08:32 |
| Qianbiao | This issue is about supporting multiple kind hardware in same server. | 08:34 |
| arne_wiebalck | Correct. | 08:34 |
| arne_wiebalck | This is what I meant with "a hw manager is supposed to manage specific hardware and it seems the intel one is not the correct one for the nodes you're trying to manage". | 08:34 |
| Qianbiao | Yes | 08:34 |
| arne_wiebalck | Does your hardware have a CNA network card? | 08:35 |
| Qianbiao | Yes | 08:35 |
| Qianbiao | and it has mlnx ib card too.. | 08:35 |
| Qianbiao | Then bug come out. | 08:36 |
| arne_wiebalck | :) | 08:36 |
| arne_wiebalck | I am not sure this is a bug. You simply do not have a hardware manager which is able to manage your hardware. | 08:36 |
| arne_wiebalck | It is a combination of two, actually. | 08:36 |
| Qianbiao | arne_wiebalck do not agree with u. InterCna mgr should not inherit generic hw mgr. | 08:38 |
| Qianbiao | currently, it makes cna manager a high priority "generic" hardware manager. | 08:38 |
| Qianbiao | none other hardware manager will work in this situation. | 08:39 |
| arne_wiebalck | this is how it is supposed to be I think | 08:39 |
| arne_wiebalck | the generic one is the most generic, it has the lowest priority | 08:39 |
| Qianbiao | a spec hardware manager should only care about its special hardware part. but not all. | 08:39 |
| arne_wiebalck | correct | 08:40 |
| Qianbiao | Current, intel cna handles all hardware. | 08:40 |
| *** derekh has joined #openstack-ironic | 08:40 | |
| Qianbiao | change "class IntelCnaHardwareManager(hardware.GenericHardwareManager):" -> "class IntelCnaHardwareManager(hardware.HardwareManager):" | 08:41 |
| Qianbiao | everything fixes. | 08:41 |
| arne_wiebalck | unless you want to run sth from the generic hardware manager | 08:41 |
| Qianbiao | yes, all none spec hardware should be handled by generic one. | 08:42 |
| arne_wiebalck | right, but will it if you don't inherit from it? | 08:42 |
| Qianbiao | so, if cna can not handle something(not implement function), it will auto use generic finally. | 08:42 |
| Qianbiao | yes | 08:42 |
| Qianbiao | if cna can not handle it, it will try low priority manager | 08:43 |
| Qianbiao | until if find one can handle it. | 08:43 |
| Qianbiao | This makes it support multiple hardware. | 08:43 |
| arne_wiebalck | oh, I think I see what you mean | 08:44 |
| arne_wiebalck | "Custom HardwareManagers should subclass hardware.HardwareManager or hardware.GenericHardwareManager. " | 08:45 |
| Qianbiao | :) | 08:45 |
| arne_wiebalck | So, the inheritance from the generic one brings functionality to the cna one which should be taken by the mlnx one. | 08:46 |
| Qianbiao | should never inherit GenericHardwareMgr. It will make the whole mechanism fail. | 08:46 |
| Qianbiao | Customer MGR | 08:46 |
| arne_wiebalck | Ok ... but why would you ever inherit from the generic one? | 08:47 |
| Qianbiao | The Intel CNA did this.. and it break things. | 08:47 |
| Qianbiao | mlnx one did the right thing. | 08:48 |
| arne_wiebalck | The docs say, one can do this. If our understanding is correct, why would you ever inherit from the generic one? | 08:48 |
| arne_wiebalck | I think we need someone like dtantsur|afk to shed some light :) | 08:49 |
| Qianbiao | :) | 08:49 |
| Qianbiao | yes | 08:49 |
| Qianbiao | I do not why, IMHO, one should never inherit generic mgr. | 08:49 |
| arne_wiebalck | exactly | 08:51 |
| arne_wiebalck | it works in almost all cases as few people have a hierarchy of manageres | 08:52 |
| arne_wiebalck | managers | 08:52 |
| arne_wiebalck | but, yeah, with our current understanding I agree with you | 08:52 |
| Qianbiao | this bug only happens when mlnx and cna card both appear in one server. | 08:53 |
| *** hjensas|afk is now known as hjensas | 08:54 | |
| arne_wiebalck | yes, but will appear with any combination of the cna manager and some other hardware | 08:55 |
| *** JohnMice has quit IRC | 08:56 | |
| Qianbiao | right. | 08:59 |
| Qianbiao | Hi, iurygregory. will this kind of patch will be accept on stable branches? | 08:59 |
| Qianbiao | https://storyboard.openstack.org/#!/story/2008144 | 08:59 |
| *** dtantsur|afk is now known as dtantsur | 09:00 | |
| dtantsur | morning ironic | 09:00 |
| Qianbiao | hey dtantsur | 09:01 |
| dtantsur | Qianbiao: as long as you don't *require* a new version, it's fine in my book | 09:01 |
| dtantsur | and hi :) | 09:01 |
| Qianbiao | o/ | 09:01 |
| dtantsur | of course assuming you won't rewrite half of the driver to support newer versions or introduce new user-visible features | 09:02 |
| Qianbiao | :) sure | 09:02 |
| dtantsur | Qianbiao: oh, good catch re inheritance of hardware managers | 09:02 |
| Qianbiao | I still remember the issue cause by me. | 09:03 |
| Qianbiao | dtantsur You agree? change the inherit class of intel cna | 09:03 |
| dtantsur | Qianbiao: I'm reading the code, it's a bit more subtle. gimme a minute | 09:04 |
| Qianbiao | sure | 09:04 |
| dtantsur | Qianbiao: yes, you're right, it has to be changed. but it's not a simple one-line change because their collect_lldp_data method calls "super", which will break | 09:05 |
| dtantsur | so GenericHardwareManager.collect_lldp_data will have to be extracted to a common place | 09:06 |
| Qianbiao | ok | 09:06 |
| Qianbiao | I will submit a patch to fix it. | 09:07 |
| dtantsur | thank you! | 09:07 |
| Qianbiao | yw | 09:07 |
| arne_wiebalck | dtantsur: we were wondering if there is a situation where one would ever inherit from the generic one ...? | 09:08 |
| arne_wiebalck | dtantsur: it works if you have only one more specific one, but with multiple managers? | 09:10 |
| dtantsur | arne_wiebalck: the problem is this line https://opendev.org/openstack/ironic-python-agent/src/branch/master/ironic_python_agent/hardware_managers/cna.py#L79 | 09:10 |
| dtantsur | it results in a copy of GenericHardwareManager with elevated priority | 09:11 |
| dtantsur | (MAINLINE vs GENERIC) | 09:11 |
| dtantsur | so it overrides everything of the same priority, including the mlnx hardware manager | 09:11 |
| dtantsur | (because GenericHardwareManager implements all methods) | 09:11 |
| arne_wiebalck | right | 09:12 |
| arne_wiebalck | but | 09:12 |
| arne_wiebalck | if you have a custom hw manager with whatever prio and you inherit from the generic one you will always have this situation | 09:12 |
| dtantsur | that's why you don't inherit unless you literally intend to raise priorities of all methods | 09:12 |
| Qianbiao | arne_wiebalck IMHO, i do not think any mgr should inherit generic one. If it inherit generic, it will make itself the generic one. | 09:12 |
| arne_wiebalck | dtantsur: yes | 09:13 |
| arne_wiebalck | Qianbiao: yes | 09:13 |
| arne_wiebalck | but the docs should probably explain this point | 09:13 |
| arne_wiebalck | "Custom HardwareManagers should subclass hardware.HardwareManager or hardware.GenericHardwareManager." | 09:13 |
| dtantsur | Qianbiao: I have a thought of a slightly different implementation | 09:14 |
| arne_wiebalck | Alternatively, should custom hw manager *ever* inherit from the generic one? | 09:14 |
| dtantsur | maybe their collect_lldp_data should call _disable_embedded_lldp_agent_in_cna_card then raise IncompatibleHardwareMethodError to fall through | 09:14 |
| dtantsur | arne_wiebalck: unlikely | 09:15 |
| dtantsur | Qianbiao: this would be an easier implementation and it would work correctly if another hardware manager also implements collect_lldp_info | 09:15 |
| Qianbiao | dtantsur this can do what we want. but i think it's not so directly. | 09:15 |
| Qianbiao | a bit wierd to me. | 09:16 |
| dtantsur | it is weird, but that's how hardware managers are design | 09:16 |
| Qianbiao | think about if there are another mgr which implement collect_lldp_info | 09:16 |
| dtantsur | IncompatbleHardwareMethodError means "go find another hardware manager to do this job" | 09:16 |
| dtantsur | (and we do have one - GenericHardwareManager) | 09:17 |
| Qianbiao | and has a priority lower than intel but great than generic | 09:17 |
| Qianbiao | then thing will go mess | 09:17 |
| dtantsur | why? | 09:17 |
| dtantsur | (btw there is no priorities in between) | 09:17 |
| Qianbiao | i see, i mean assume. | 09:18 |
| Qianbiao | then generic mgr's collect_lldp_info may not be called, but the one between intel and generic | 09:18 |
| Qianbiao | this is not what we want | 09:18 |
| dtantsur | why not? | 09:18 |
| dtantsur | I think this is exactly what we want, because it may be able to handle some other network adapters | 09:19 |
| dtantsur | (collect_lldp_info interface is misdesigned, but that's another topic) | 09:19 |
| Qianbiao | hmm. I do not know how to explain. ~~ | 09:20 |
| Qianbiao | because intel cna collect lldp exactly wants "1. disable xxx, 2. collect lldp exactly as the generic one did now" | 09:20 |
| Qianbiao | if we raise error, the part2 is not so sure. | 09:21 |
| dtantsur | I don't think part 2 is correct, I think it's an oversight from whoever wrote that | 09:21 |
| dtantsur | because there may be other vendors on the same machine | 09:21 |
| Qianbiao | IN this sight, you are right. | 09:21 |
| Qianbiao | this depends on how intel cna want the "collect lldp" looks like. | 09:22 |
| dtantsur | it seems that all they want is to disable some built-in LLDP facility beforehand | 09:23 |
| dtantsur | I'm not sure why they don't do it on start-up, to be honest | 09:23 |
| Qianbiao | ok, will patch it as you said. | 09:24 |
| *** Lucas_Gray has joined #openstack-ironic | 09:25 | |
| Qianbiao | But if so, every spec hardware mgr need to consider whether there are other hardware mgr will effect current one? | 09:26 |
| Qianbiao | The 3rd hardware mgr should finish all it want to do, but not leave it to generic one or others. | 09:26 |
| openstackgerrit | Ankit Kumar proposed openstack/ironic master: Adding changes for iso less vmedia support https://review.opendev.org/752001 | 09:27 |
| dtantsur | Qianbiao: that's a common problem with hardware managers: they don't operate in isolation | 09:28 |
| Qianbiao | ok make sense too. | 09:31 |
| iurygregory | yay IPE installed with bifrost \o/ now I just need to check if I can collect data =D | 09:32 |
| *** JamesBenson has joined #openstack-ironic | 09:33 | |
| openstackgerrit | Arne Wiebalck proposed openstack/ironic-python-agent master: [doc] Clarify custom hardware manager inheritance https://review.opendev.org/752002 | 09:35 |
| *** JamesBenson has quit IRC | 09:37 | |
| *** yolanda has joined #openstack-ironic | 09:37 | |
| openstackgerrit | QianBiao Ng proposed openstack/ironic stable/ussuri: opt: Enhance old stable branches to use latest python-ibmcclient https://review.opendev.org/752006 | 09:40 |
| *** tkajinam has quit IRC | 09:49 | |
| Qianbiao | dtantsur i got another issue. The inspect ramdisk log is always truncated. | 10:18 |
| Qianbiao | any idea what may cause this. | 10:19 |
| dtantsur | I remember solving a similar problem in RDO ramdisks, I can take a look after lunch | 10:29 |
| openstackgerrit | QianBiao Ng proposed openstack/ironic stable/ussuri: opt: Enhance old stable branches to use latest python-ibmcclient https://review.opendev.org/752006 | 10:37 |
| uzumaki | quick question, if I have a stand alone ironic setup, and want to use the openstack CLI to interact with it, how do I setup the RC file? | 10:37 |
| iurygregory | uzumaki, you can setup a clouds.yaml file fore it | 10:38 |
| iurygregory | for* | 10:38 |
| uzumaki | and I'll have to mount that file to the ironic pod then? | 10:39 |
| iurygregory | are you talking about metal3? | 10:40 |
| uzumaki | i am, yes | 10:41 |
| iurygregory | if you are deployment with the dev-env it should create the file for you | 10:41 |
| uzumaki | I got a messed up clean failed, and I don't want to scroll through gazillion lines of conductor logs, what I see in the node as "last_error" is timeout reached while waiting for ramdisk, I supposed I'll clear the maintenance flag and get rid of the node | 10:41 |
| iurygregory | you just need to say export OS_CLOUD=metal3 | 10:41 |
| uzumaki | I setup the environment, trying to be smart SMH xD | 10:42 |
| iurygregory | http://paste.openstack.org/show/797873/ | 10:42 |
| iurygregory | you can put a file like this in the dir you need and export OS_CLOUD=<name> | 10:43 |
| iurygregory | and you should be able to use the openstack cli | 10:43 |
| uzumaki | mucho gracias! | 10:43 |
| *** ociuhandu has joined #openstack-ironic | 10:58 | |
| *** ociuhandu_ has joined #openstack-ironic | 11:00 | |
| *** JamesBenson has joined #openstack-ironic | 11:02 | |
| *** ociuhandu has quit IRC | 11:03 | |
| uzumaki | no connection adapters were found? what's that supposed to mean? iurygregory | 11:07 |
| iurygregory | uzumaki, depends your clouds.yml is correct and did you set the env OS_CLOUD? | 11:07 |
| uzumaki | I did | 11:07 |
| uzumaki | does it need a /v1/ or a /v1 only? | 11:08 |
| uzumaki | the endpoints | 11:08 |
| iurygregory | http://<ip>:6385 | 11:09 |
| iurygregory | http://<ip>:5050 | 11:09 |
| uzumaki | ah! it worked | 11:09 |
| uzumaki | I was missing the http:// SMH | 11:09 |
| uzumaki | merci | 11:09 |
| iurygregory | =) glad it worked | 11:11 |
| uzumaki | \o/ what's sept 28? I recall yesterday you talked about an interesting CZ holiday | 11:11 |
| uzumaki | I'm curious | 11:11 |
| iurygregory | its a holiday thats all I know lol | 11:11 |
| uzumaki | because that's all that mtters? xD | 11:11 |
| iurygregory | and I will be moving my things to a new flat | 11:11 |
| uzumaki | Ohhh une apartment nouvelle? felicitations! | 11:12 |
| iurygregory | what? O.o | 11:12 |
| uzumaki | Sorry, I got so excited I went French :D | 11:12 |
| iurygregory | french? | 11:12 |
| iurygregory | oh ok | 11:12 |
| iurygregory | =) | 11:12 |
| uzumaki | :D congratulations! \o/ | 11:13 |
| iurygregory | ty | 11:13 |
| uzumaki | how much stuff do you have to carry? | 11:13 |
| *** zzzeek has quit IRC | 11:24 | |
| iurygregory | a lot hehehe | 11:24 |
| iurygregory | going to rent a van | 11:24 |
| *** zzzeek has joined #openstack-ironic | 11:24 | |
| * iurygregory lunch time | 11:25 | |
| Qianbiao | dtantsur thanks | 11:38 |
| *** Goneri has joined #openstack-ironic | 11:44 | |
| *** rh-jelabarre has joined #openstack-ironic | 11:52 | |
| *** mkrai has quit IRC | 11:52 | |
| *** mkrai_ has joined #openstack-ironic | 11:53 | |
| *** rh-jelabarre has quit IRC | 11:54 | |
| *** jtomasek has quit IRC | 11:54 | |
| *** rh-jelabarre has joined #openstack-ironic | 11:54 | |
| openstackgerrit | Ankit Kumar proposed openstack/ironic master: Adding changes for iso less vmedia support https://review.opendev.org/752001 | 11:56 |
| openstackgerrit | QianBiao Ng proposed openstack/ironic-python-agent master: Fix: make Intel CNA hardware manager none generic https://review.opendev.org/752024 | 11:56 |
| *** ociuhandu_ has quit IRC | 12:01 | |
| *** ociuhandu has joined #openstack-ironic | 12:01 | |
| dtantsur | has anyone looked into what is happening with the CI? I see quite a few failures | 12:12 |
| iurygregory | dtantsur, I saw failures related to oslo.serialization version not found | 12:13 |
| iurygregory | probably some mirror problems.. | 12:14 |
| dtantsur | yeah, that's likely a mirror lag | 12:14 |
| iurygregory | maybe they have something on #opendev | 12:14 |
| *** Lucas_Gray has quit IRC | 12:14 | |
| dtantsur | on #openstack-infra | 12:14 |
| dtantsur | it seems to be an upstream pypi mirror problem | 12:14 |
| dtantsur | great timing, sigh | 12:15 |
| iurygregory | ouch =( | 12:15 |
| iurygregory | sometimes I just hate systemd... | 12:15 |
| *** Lucas_Gray has joined #openstack-ironic | 12:16 | |
| iurygregory | the app run if put the commands on shell, on systemd Active: inactive (dead) -.-' | 12:16 |
| *** Goneri has quit IRC | 12:23 | |
| *** sshnaidm has quit IRC | 12:25 | |
| dtantsur | make sure you run in foreground | 12:26 |
| iurygregory | dtantsur, http://paste.openstack.org/show/797876/ my service looks like this (In devstack it works).. | 12:28 |
| *** ociuhandu has quit IRC | 12:28 | |
| dtantsur | iurygregory: other than FLASK_DEBUG=1, looks okay to me.. | 12:29 |
| dtantsur | iurygregory: oh, and please don't use log files, log to journald instead | 12:30 |
| dtantsur | (that's maybe why you don't see any errors) | 12:30 |
| iurygregory | the logs are empty .-. thats the strange part | 12:31 |
| iurygregory | + Active: inactive (dead) | 12:31 |
| iurygregory | http://paste.openstack.org/show/797877/ | 12:31 |
| dtantsur | what exactly does the -D flag do? | 12:31 |
| dtantsur | I don't see it used on sushy-tools | 12:32 |
| iurygregory | daemon | 12:32 |
| iurygregory | in the devstack setup we use like this (for ipe) | 12:32 |
| iurygregory | https://github.com/openstack/ironic-prometheus-exporter/blob/master/devstack/plugin.sh#L36 | 12:32 |
| dtantsur | iurygregory: yep, exactly this thing you must not use with systemd :) | 12:32 |
| dtantsur | systemd managed processes should run in foreground, unless specially configured to talk to systemd | 12:33 |
| dtantsur | I think it just forks away and systemd loses track of it | 12:33 |
| iurygregory | hummm | 12:34 |
| * iurygregory removes -D | 12:34 | |
| janders | https://i.imgur.com/D1XK8nk.gif | 12:34 |
| iurygregory | Active: active (running) =O | 12:35 |
| iurygregory | oh yes! | 12:35 |
| iurygregory | https://media.giphy.com/media/lOa0tPKiMLdqVdFiS8/giphy.gif | 12:35 |
| dtantsur | we should probably use `--access-logfile -` to make it log to stderr | 12:35 |
| dtantsur | (--error-logfile defaults to - as it seems) | 12:35 |
| openstackgerrit | Dmitry Tantsur proposed openstack/bifrost master: redfish-emulator: log handled requests https://review.opendev.org/752036 | 12:38 |
| *** priteau has quit IRC | 12:42 | |
| *** Goneri has joined #openstack-ironic | 12:49 | |
| *** sshnaidm has joined #openstack-ironic | 12:50 | |
| *** priteau has joined #openstack-ironic | 12:55 | |
| *** uzumaki has quit IRC | 12:59 | |
| janders | see you tomorrow Ironic o/ | 13:03 |
| *** ociuhandu has joined #openstack-ironic | 13:12 | |
| *** priteau has quit IRC | 13:15 | |
| TheJulia | good morning | 13:17 |
| openstackgerrit | Iury Gregory Melo Ferreira proposed openstack/bifrost master: Add ipe installation to bifrost https://review.opendev.org/752042 | 13:19 |
| TheJulia | is CI any happier today? | 13:21 |
| TheJulia | nope | 13:24 |
| *** ociuhandu has quit IRC | 13:29 | |
| dtantsur | morning TheJulia | 13:32 |
| dtantsur | no, the CI is in a terrible state | 13:32 |
| TheJulia | :( | 13:35 |
| TheJulia | all the fastly proxies? | 13:35 |
| dtantsur | TheJulia: seems to boil down to them, yes | 13:36 |
| openstackgerrit | vinay kumar muddu proposed openstack/ironic master: Adds ilo-uefi-https boot interface to ilo5 https://review.opendev.org/745501 | 13:38 |
| openstackgerrit | Yogesh Ramachandra proposed openstack/ironic master: Fix idrac-wsman bios factory-reset setting https://review.opendev.org/748696 | 13:39 |
| *** ociuhandu has joined #openstack-ironic | 13:46 | |
| *** tzumainn has joined #openstack-ironic | 13:47 | |
| TheJulia | well, code reviews could be a thing instead! | 13:48 |
| *** uzumaki has joined #openstack-ironic | 13:48 | |
| *** yolanda has quit IRC | 13:53 | |
| *** yolanda has joined #openstack-ironic | 13:54 | |
| *** hjensas is now known as hjensas|afk | 13:56 | |
| *** bbezak_ has joined #openstack-ironic | 13:57 | |
| *** sri___ has joined #openstack-ironic | 13:57 | |
| *** arne_wiebalck_ has joined #openstack-ironic | 13:57 | |
| *** andrein_ has joined #openstack-ironic | 13:57 | |
| *** phrobb_ has joined #openstack-ironic | 13:57 | |
| *** v1k0d3n_ has joined #openstack-ironic | 13:57 | |
| *** gmann_ has joined #openstack-ironic | 13:57 | |
| *** guilhermesp_ has joined #openstack-ironic | 13:57 | |
| *** priteau has joined #openstack-ironic | 13:57 | |
| *** sri_ has quit IRC | 14:00 | |
| *** guilhermesp has quit IRC | 14:00 | |
| *** bbezak has quit IRC | 14:00 | |
| *** v1k0d3n has quit IRC | 14:00 | |
| *** fungi has quit IRC | 14:00 | |
| *** arne_wiebalck has quit IRC | 14:00 | |
| *** andrein has quit IRC | 14:00 | |
| *** seba has quit IRC | 14:00 | |
| *** logan- has quit IRC | 14:00 | |
| *** gmann has quit IRC | 14:00 | |
| *** phrobb has quit IRC | 14:00 | |
| *** jroll has quit IRC | 14:00 | |
| *** dtantsur has quit IRC | 14:00 | |
| *** antonym has quit IRC | 14:00 | |
| *** sri___ is now known as sri_ | 14:00 | |
| *** bbezak_ is now known as bbezak | 14:00 | |
| *** arne_wiebalck_ is now known as arne_wiebalck | 14:00 | |
| *** guilhermesp_ is now known as guilhermesp | 14:00 | |
| *** v1k0d3n_ is now known as v1k0d3n | 14:00 | |
| *** logan_ has joined #openstack-ironic | 14:00 | |
| *** phrobb_ is now known as phrobb | 14:00 | |
| *** gmann_ is now known as gmann | 14:00 | |
| *** andrein_ is now known as andrein | 14:00 | |
| *** logan_ is now known as logan- | 14:00 | |
| *** jamesdenton has quit IRC | 14:03 | |
| *** jamesdenton has joined #openstack-ironic | 14:05 | |
| *** antonym has joined #openstack-ironic | 14:07 | |
| *** jroll has joined #openstack-ironic | 14:07 | |
| *** fungi has joined #openstack-ironic | 14:08 | |
| *** dtantsur has joined #openstack-ironic | 14:08 | |
| *** rloo has joined #openstack-ironic | 14:15 | |
| dtantsur | TheJulia: do you know whom to ask to allow changing timezone for the openstack summit schedule? | 14:15 |
| dtantsur | it's not exactly convenient to do the math on fly | 14:15 |
| dtantsur | (although with a virtual summit I can watch everything later) | 14:17 |
| TheJulia | dtantsur: jamesmcarthur (jimmy) would be the person to talk to | 14:21 |
| dtantsur | TheJulia: thanks! I will unless this thought will get buried under dozens of other thoughts.. | 14:22 |
| TheJulia | :) | 14:28 |
| TheJulia | awesome | 14:28 |
| TheJulia | iurygregory: w/r/t https://review.opendev.org/#/c/705030/ i think we need to make it non-voting for the multinode job until we sort out the issues there. | 14:29 |
| patchbot | patch 705030 - ironic - Native zuulv3 grenade multinode multitenant - 33 patch sets | 14:29 |
| TheJulia | it presently votes in the check run to fail which will prevent it from merging. | 14:29 |
| openstackgerrit | QianBiao Ng proposed openstack/ironic-python-agent master: Fix: make Intel CNA hardware manager none generic https://review.opendev.org/752024 | 14:31 |
| openstackgerrit | vinay kumar muddu proposed openstack/ironic master: Adds ilo-uefi-https boot interface to ilo5 https://review.opendev.org/745501 | 14:31 |
| openstackgerrit | Dmitry Tantsur proposed openstack/ironic master: Deprecate the iscsi deploy interface https://review.opendev.org/750204 | 15:02 |
| openstackgerrit | QianBiao Ng proposed openstack/ironic-python-agent stable/train: Fix: make Intel CNA hardware manager none generic https://review.opendev.org/752065 | 15:07 |
| TheJulia | Qianbiao: typically we only backport once the change has merged to master | 15:09 |
| TheJulia | Qianbiao: but then we need to backport that to... multiple branches :) | 15:09 |
| Qianbiao | TheJulia I just need to click cherry-pick and choose branches, right? | 15:10 |
| Qianbiao | all branch? | 15:10 |
| *** dhellmann_ has joined #openstack-ironic | 15:10 | |
| *** dhellmann has quit IRC | 15:11 | |
| *** dhellmann_ is now known as dhellmann | 15:12 | |
| TheJulia | no. In order typically master -> the bugfix branches -> stable/ussuri, stable/train, stable/stein, stable/rocky (optional, but preferred), stable/queens (optional, but preferred) | 15:12 |
| Qianbiao | ok | 15:13 |
| tosky | not sure whether clicking from the UI does the same as git cherry-pick -x | 15:13 |
| Qianbiao | I need to wait it merged to master? after that i can cherry pick. | 15:14 |
| TheJulia | if merged, the ui does the proper tagging of the commit message | 15:14 |
| TheJulia | but *shrug* | 15:14 |
| TheJulia | as long as we know where it came from in master I think we're good | 15:14 |
| TheJulia | Qianbiao: basically yes | 15:15 |
| Qianbiao | ok, np, i will do that. | 15:15 |
| TheJulia | awesome, thanks! | 15:15 |
| *** ociuhandu_ has joined #openstack-ironic | 15:17 | |
| Qianbiao | yw :) | 15:18 |
| *** ociuhandu has quit IRC | 15:21 | |
| *** ociuhandu_ has quit IRC | 15:22 | |
| *** mkrai_ has quit IRC | 15:23 | |
| Qianbiao | dtantsur the truncated inspect ramdisk log issue, should i create a story to track it | 15:27 |
| dtantsur | yes please (against IPA-builder) | 15:27 |
| Qianbiao | It can be reproducted in my env. Not sure why. | 15:27 |
| Qianbiao | ok | 15:27 |
| Qianbiao | dtantsur https://storyboard.openstack.org/#!/story/2008148 | 15:35 |
| *** hjensas|afk is now known as hjensas | 15:35 | |
| *** ianychoi__ has joined #openstack-ironic | 15:41 | |
| *** ianychoi_ has quit IRC | 15:45 | |
| TheJulia | JayF: Yeah, the change from poll to eventlet for the heartbeat is definitely breaking :\ | 15:47 |
| TheJulia | err, select | 15:48 |
| *** uzumaki has quit IRC | 15:49 | |
| TheJulia | not sure why but requests may not work it seems | 15:50 |
| dtantsur | maybe the sleep change? | 15:53 |
| TheJulia | If I'm remembering how requests calls all the way into urllib... I don't think it should but maybe it is a timing thing at that point | 15:54 |
| TheJulia | and then becomes random hit/miss | 15:54 |
| TheJulia | which is actually kind of like what we're seeing | 15:54 |
| TheJulia | but... yeah. | 15:54 |
| JayF | Yeah; I'm not sure, and bluntly I'm not sure how much value it brings to troubleshoot that. | 15:56 |
| JayF | I'd be happy to abandon it if you all agree. | 15:56 |
| TheJulia | I was kind of wondering the same question | 15:56 |
| TheJulia | I mean... I love how much cleaner it is | 15:56 |
| dtantsur | can we try with reverting to sleep(0.1) first? | 15:56 |
| dtantsur | at least it would give us a hint | 15:57 |
| openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent master: Simplify heartbeating by removing use of select() https://review.opendev.org/746774 | 15:58 |
| JayF | I mean, I'm skeptical given sleep(0) wasn't breaking before | 15:58 |
| JayF | but it's a simple enough edit | 15:58 |
| dtantsur | yeah, I doubt as well.. but to rule out for sure | 15:58 |
| JayF | I suspect this is yet-another-case of needing real threading OR using eventlet everywhere, instead of this wacky middle ground | 15:58 |
| dtantsur | agree | 15:58 |
| dtantsur | time for futurist and its periodic tasks? | 15:59 |
| TheJulia | definitely | 15:59 |
| * TheJulia might try to convince someone on that | 16:00 | |
| dtantsur | JayF: another thought: what if we import eventlet.event.Event explicitly? instead of threading.Event? | 16:00 |
| JayF | I like that idea, a lot actually | 16:01 |
| JayF | lets see how the first change tests out | 16:01 |
| TheJulia | ++ | 16:01 |
| JayF | I'll try to remember to try that after | 16:01 |
| JayF | but if I forget feel free to edit in (your) tomorrow morning | 16:01 |
| TheJulia | if it is the sleep in general, then importing and using directly will cause the heartbeater to most likely freezze | 16:01 |
| *** mkrai has joined #openstack-ironic | 16:01 | |
| TheJulia | I _bet_ this is exactly what is happening with requests but without injecting a bunch of debugging in to requests I don't think we would see it | 16:01 |
| *** lucasagomes has quit IRC | 16:01 | |
| dtantsur | and probably use eventlet.spawn instead of threading.Thread | 16:02 |
| dtantsur | (we can play with combinations of these solutions) | 16:02 |
| JayF | ...or just monkey patch thread, too | 16:02 |
| JayF | if it didn't kaboom, it'd get us closer to "eventlet everything" | 16:02 |
| dtantsur | that's probably the first option to try, really | 16:04 |
| TheJulia | I have indecisive thoughts on this matter that request tequila to be delivered | 16:06 |
| *** gyee has joined #openstack-ironic | 16:06 | |
| dtantsur | python-eventlet.spec: Requires: tequila>=2 | 16:07 |
| * dtantsur wonders if people understand rpm jokes | 16:07 | |
| openstackgerrit | Verification of a change to openstack/ironic failed: Also wipe agent token on manual power off or reboot https://review.opendev.org/751813 | 16:08 |
| TheJulia | I wonder if stevebaker would be interested in going down the rabbit hole of looking at evenleting it completely | 16:09 |
| TheJulia | dtantsur: oh yes, I got it :) | 16:09 |
| JayF | dtantsur: they make my head spin | 16:09 |
| JayF | (bam, double-rpm joke) | 16:09 |
| dtantsur | :D | 16:10 |
| -openstackstatus- NOTICE: Our PyPI caching proxies are serving stale package indexes for some packages. We think because PyPI's CDN is serving stale package indexes. We are sorting out how we can either fix or workaround that. In the meantime updating requirements is likely the wrong option. | 16:10 | |
| *** mkrai has quit IRC | 16:10 | |
| TheJulia | all my concern and worry is largely around requests timeouts since we've seen the agent tends to get fired up in super grumpy networks | 16:11 |
| TheJulia | speaking of which... inspection connection retries! | 16:11 |
| Qianbiao | hello, when package ipa with dib, where ipa code will be put? | 16:19 |
| Qianbiao | which path | 16:19 |
| dtantsur | Qianbiao: installed into /opt/ironic-python-agent, I think | 16:20 |
| dtantsur | https://opendev.org/openstack/ironic-python-agent-builder/src/branch/master/dib/ironic-python-agent-ramdisk/install.d/ironic-python-agent-ramdisk-source-install/60-ironic-python-agent-ramdisk-install#L12-L14 | 16:21 |
| Qianbiao | <dtantsur> thanks | 16:22 |
| Qianbiao | .... it seems it does not package latest code. | 16:23 |
| TheJulia | What version is getting packaged? | 16:23 |
| Qianbiao | TheJulia I am using this script: https://storyboard.openstack.org/#!/story/2008148 | 16:24 |
| Qianbiao | it should use my local repo fold HEAD commit | 16:24 |
| Qianbiao | but it does not. | 16:24 |
| dtantsur | Qianbiao: how exactly do you configure that? | 16:25 |
| Qianbiao | check the story: export DIB_REPOLOCATION_ironic_python_agent="/opt/stack/ironic-python-agent" | 16:25 |
| Qianbiao | dtantsur ^^ | 16:26 |
| dtantsur | Qianbiao: do you set DIB_REPOREF_ironic_python_agent=HEAD? I'm not 100% sure it's needed, but we do it in the CI. | 16:27 |
| Qianbiao | yes | 16:27 |
| Qianbiao | i set that. | 16:27 |
| dtantsur | aha, I see now | 16:27 |
| dtantsur | hmmm, the instruction on the story looks correct | 16:29 |
| TheJulia | I'm thinking the same | 16:29 |
| TheJulia | do you have the build log from the ramdisk creation? | 16:29 |
| TheJulia | I ask because that version of the agent is not logging anything upon startup | 16:29 |
| TheJulia | it could be your local branch, but we've had that in the agent for close to a year I think. | 16:29 |
| Qianbiao | <TheJulia> will rebuild for now | 16:30 |
| Qianbiao | * --for-- | 16:30 |
| dtantsur | as to trimmed logging, I think we need something like https://review.rdoproject.org/r/#/c/21142/ | 16:31 |
| Qianbiao | how can i get the ipa versin in built initramfs. | 16:32 |
| Qianbiao | i have uncompress it. | 16:32 |
| *** jlvillal has quit IRC | 16:32 | |
| *** jlvillal has joined #openstack-ironic | 16:32 | |
| Qianbiao | <dtantsur> you mean log to systemd, and then journal get it, then send it back. | 16:33 |
| dtantsur | well, yeah. currently we log to stdout, which gets to journald, which we send back | 16:33 |
| dtantsur | this stdout -> journald process is buffered, so something is usually trimmed | 16:33 |
| dtantsur | the RDO change I made some time ago makes IPA (or rather oslo.log) talk to journald directly | 16:34 |
| dtantsur | which is unbuffered and real-time | 16:34 |
| Qianbiao | nice | 16:35 |
| Qianbiao | the version is: ironic_python_agent-5.0.4.dev8-py3.6.egg-info | 16:36 |
| dtantsur | bfournie: could you check if https://review.opendev.org/751813 fixes https://storyboard.openstack.org/#!/story/2008097 ? | 16:37 |
| patchbot | patch 751813 - ironic - Also wipe agent token on manual power off or reboot - 2 patch sets | 16:37 |
| TheJulia | from with-in the runtime is there not a way to turn off python's buffered io? | 16:38 |
| dtantsur | TheJulia: I don't remember, I think I tried. It's not just about Python, also about glibc, I'm afraid.. | 16:40 |
| TheJulia | that would explain why only the env-var works afaik | 16:40 |
| *** dtantsur is now known as dtantsur|afk | 16:46 | |
| dtantsur|afk | o/ | 16:46 |
| dtantsur|afk | see you tomorrow, hopefully with the CI working | 16:46 |
| TheJulia | goodnight! | 16:46 |
| Qianbiao | see you | 16:47 |
| Qianbiao | TheJulia http://paste.openstack.org/show/797894/ | 16:47 |
| Qianbiao | it's wierd, this head is not my HEAD local | 16:47 |
| Qianbiao | My git log is: http://paste.openstack.org/show/797895/ | 16:49 |
| TheJulia | try setting the ref to ab639cc672d984f897ad1117c0e98d512dc0c7e6 | 16:49 |
| TheJulia | I suspect this might be a mechanics interaction with the internal clone | 16:50 |
| Qianbiao | agree | 16:51 |
| Qianbiao | will try | 16:51 |
| Qianbiao | TheJulia http://paste.openstack.org/show/797896/ | 16:54 |
| Qianbiao | it seems it must get ref from origin? | 16:55 |
| Qianbiao | or i should use a short commit id | 16:55 |
| TheJulia | git reflog | 16:55 |
| TheJulia | yeah, it really wants the ref :\ | 16:55 |
| TheJulia | but reflog should give you visibility on what the tag is to use | 16:56 |
| *** derekh has quit IRC | 16:56 | |
| *** jtomasek has joined #openstack-ironic | 16:59 | |
| Qianbiao | TheJulia it did not support local ref from not origin. http://paste.openstack.org/show/797897/ | 17:01 |
| Qianbiao | ~~ | 17:01 |
| TheJulia | ugh | 17:01 |
| openstackgerrit | QianBiao Ng proposed openstack/ironic-python-agent master: Fix: make Intel CNA hardware manager none generic https://review.opendev.org/752024 | 17:32 |
| Qianbiao | sorry, TheJulia, did not run tox testcase locally. | 17:32 |
| Qianbiao | (network here is really hard to download packages with tox) | 17:33 |
| *** tosky has quit IRC | 17:40 | |
| *** Qianbiao is now known as Qianbiao|afk | 17:44 | |
| *** Qianbiao|afk has quit IRC | 17:44 | |
| openstackgerrit | QianBiao Ng proposed openstack/ironic stable/ussuri: opt: Enhance old stable branches to use latest python-ibmcclient https://review.opendev.org/752006 | 17:46 |
| *** Lucas_Gray has quit IRC | 17:50 | |
| *** dsneddon has quit IRC | 17:53 | |
| *** cdearborn has joined #openstack-ironic | 18:20 | |
| *** ociuhandu has joined #openstack-ironic | 18:29 | |
| *** ociuhandu has quit IRC | 18:34 | |
| *** VisBits has joined #openstack-ironic | 18:35 | |
| VisBits | Hi, I am curious how the driver implementation works for adding the PXE options to neutron-dhcp-agent.. i cannot figure how how that works. My issue is i have a neutron network (not discovery or cleaning) that is adding the PXE entries for the ironic pxe server and my baremetal host are rebooting into discovery upon provision.. | 18:35 |
| *** zzzeek has quit IRC | 18:39 | |
| *** zzzeek has joined #openstack-ironic | 18:41 | |
| TheJulia | o/ | 18:44 |
| TheJulia | VisBits: so that sounds like your ironic-inspector service is not filtering its configuration or firewalling properly | 18:45 |
| TheJulia | well | 18:46 |
| TheJulia | are the PXE entries on the port themselves? | 18:46 |
| TheJulia | the settings would be set if the machine is set to network boot. We recommend local booting machines. | 18:46 |
| *** VisBits has quit IRC | 18:48 | |
| TheJulia | \o/ | 18:49 |
| *** zzzeek has quit IRC | 18:51 | |
| *** VisBits has joined #openstack-ironic | 18:52 | |
| *** Lucas_Gray has joined #openstack-ironic | 18:52 | |
| *** zzzeek has joined #openstack-ironic | 18:53 | |
| openstackgerrit | Julia Kreger proposed openstack/ironic-python-agent master: Fix: make Intel CNA hardware manager none generic https://review.opendev.org/752024 | 18:53 |
| *** zzzeek has quit IRC | 19:00 | |
| *** zzzeek has joined #openstack-ironic | 19:03 | |
| *** zzzeek has quit IRC | 19:10 | |
| *** zzzeek has joined #openstack-ironic | 19:12 | |
| *** priteau has quit IRC | 19:16 | |
| *** dsneddon has joined #openstack-ironic | 19:26 | |
| openstackgerrit | Christopher Dearborn proposed openstack/ironic master: Redfish driver firmware update https://review.opendev.org/749619 | 19:27 |
| *** rloo has quit IRC | 19:40 | |
| *** rloo has joined #openstack-ironic | 19:40 | |
| openstackgerrit | Verification of a change to openstack/ironic-inspector failed: Add NVIDIA V100 to known accelerator devices https://review.opendev.org/751660 | 20:29 |
| *** stevebaker has quit IRC | 21:00 | |
| openstackgerrit | Christopher Dearborn proposed openstack/ironic master: Redfish driver firmware update https://review.opendev.org/749619 | 21:10 |
| *** stevebaker has joined #openstack-ironic | 21:23 | |
| *** JamesBenson has quit IRC | 21:24 | |
| *** Lucas_Gray has quit IRC | 21:50 | |
| *** Lucas_Gray has joined #openstack-ironic | 21:51 | |
| *** k_mouza has joined #openstack-ironic | 21:53 | |
| iurygregory | TheJulia, yeah I will update the patch to put non-voting =) | 22:09 |
| stevebaker | TheJulia: what is still not eventletted? | 22:17 |
| * stevebaker looks for where to start in the backscroll | 22:19 | |
| *** rcernin has joined #openstack-ironic | 22:25 | |
| TheJulia | iurygregory: thanks | 22:27 |
| TheJulia | stevebaker: so the tl;dr is that the agent has some... interesting code | 22:28 |
| TheJulia | stevebaker: specifcially for the heartbeater which we launch as a separate thread and only kill it off if we need to exit. That is basically unchanged and this continues to kind of bite us in other areas where we end up seeing weird things with eventlet/threading in other parts of the app | 22:29 |
| stevebaker | TheJulia: I see. When I have a gap I'll see if I can get my head around how it works currently | 22:32 |
| TheJulia | happy to spend some time walking through it with you | 22:33 |
| TheJulia | but yeah, someone who has never really looked deeply into is not indoctrinated into it either :) | 22:33 |
| TheJulia | woot, looks like CI issues may soon be resoled | 22:35 |
| *** k_mouza has quit IRC | 22:51 | |
| *** tkajinam has joined #openstack-ironic | 22:51 | |
| JayF | stevebaker: essentially: in the __init__.py where we monkey patch, change monkey patch to all=True, and image streaming breaks. That's the first thing. You'll probably also need https://review.opendev.org/#/c/746774/ | 22:55 |
| patchbot | patch 746774 - ironic-python-agent - Simplify heartbeating by removing use of select() - 25 patch sets | 22:55 |
| JayF | stevebaker: (in the gate; image streaming is the metalsmith job) | 22:56 |
| openstackgerrit | Jay Faulkner proposed openstack/ironic-python-agent master: Simplify heartbeating by removing use of select() https://review.opendev.org/746774 | 22:58 |
| JayF | ^^ sleep(0.1) helped nothing; trying the monkey_patch of threading as discussed this morning | 22:58 |
| *** k_mouza has joined #openstack-ironic | 23:00 | |
| *** Goneri has quit IRC | 23:03 | |
| *** k_mouza has quit IRC | 23:05 | |
| TheJulia | blarg, ci still not happy | 23:10 |
| TheJulia | JayF: ack :( | 23:11 |
| VisBits | Hi, I am curious how the driver implementation works for adding the PXE options to neutron-dhcp-agent.. i cannot figure how how that works. My issue is i have a neutron network (not discovery or cleaning) that is adding the PXE entries for the ironic pxe server and my baremetal host are rebooting into discovery upon provision.. | 23:22 |
| *** k_mouza has joined #openstack-ironic | 23:46 | |
| *** k_mouza has quit IRC | 23:50 | |
| *** tzumainn has quit IRC | 23:52 | |
| *** Lucas_Gray has quit IRC | 23:55 | |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
