| *** pmannidi is now known as pmannidi|brb | 01:51 | |
| *** pmannidi|brb is now known as pmannidi | 02:39 | |
| opendevreview | Steve Baker proposed openstack/ironic master: WIP Use cp to copy local images to preserve sparseness https://review.opendev.org/c/openstack/ironic/+/800152 | 03:33 |
|---|---|---|
| *** pmannidi is now known as pmannidi|Away | 05:45 | |
| *** pmannidi|Away is now known as pmannidi | 05:48 | |
| *** pmannidi is now known as pmannidi|Away | 05:48 | |
| *** pmannidi|Away is now known as pmannidi | 05:54 | |
| *** pmannidi is now known as pmannidi|Away | 05:54 | |
| *** pmannidi|Away is now known as pmannidi | 05:55 | |
| *** pmannidi is now known as pmannidi|Away | 05:55 | |
| *** pmannidi|Away is now known as pmannidi | 05:58 | |
| *** pmannidi is now known as pmannidi|Away | 05:58 | |
| *** pmannidi|Away is now known as pmannidi | 05:59 | |
| *** pmannidi is now known as pmannidi|Away | 05:59 | |
| *** pmannidi|Away is now known as pmannidi | 06:08 | |
| iurygregory | TheJulia, tks! | 06:37 |
| iurygregory | good morning janders and Ironic, happy friday | 06:37 |
| arne_wiebalck | Good morning iurygregory janders and Ironic! | 06:47 |
| iurygregory | morning arne_wiebalck o/ | 06:48 |
| jssfr | good morning! where can I find documentation on how ironic tells ironic-python-agent to do a clean? in my lab, my nodes have suddenly stopped cleaning and I have no clue why (stuck in clean wait until they eventually fail; they do netboot into IPA though) and I’d like to walk the entire chain of command until I find out whats wrong. | 06:55 |
| jssfr | first stop is the kernel commandline passed to IPA and I’d like to validate it against documentation, assuming that’s how IPA learns that it should do some cleaning. | 06:55 |
| arne_wiebalck | jssfr: the kernel only tells the IPA where to call back to, the conductor then tells the IPA what to do | 06:56 |
| arne_wiebalck | jssfr: the most likely explanation for a node stuck in cleaning is that the node is in 'maintenance' | 06:57 |
| arne_wiebalck | jssfr: so check the maintenance state with a 'baremetal node show ...' | 06:57 |
| jssfr | ohh | 07:00 |
| jssfr | so many misconceptions cleared in just a few sentences :D | 07:00 |
| * jssfr goes reading up on what "maintenance" really means | 07:00 | |
| jssfr | look at that, cleared the maintenance flag and *boom* node cleaned | 07:01 |
| jssfr | thank you very much, arne_wiebalck :) | 07:02 |
| jssfr | now I wonder how much of the "weirdness" I saw can be attributed to me not paying attention to the maintenance flag :( | 07:04 |
| arne_wiebalck | jssfr: this flag will become a good "friend" over time :-D | 07:07 |
| arne_wiebalck | zaneb: TheJulia: dtantsur: thanks for the patch and the backports of https://review.opendev.org/c/openstack/ironic-python-agent/+/788588, we just ran into this :) | 07:45 |
| *** mgoddard- is now known as mgoddard | 08:40 | |
| jssfr | oooh boy. I think I found out the hard way that my hardware is of the kind which does not respond well to `ipmitool chassis bootdev foo` without the magic `0x00 0x08 0x03 0x08` incantation before. | 10:36 |
| opendevreview | Pierre Riteau proposed openstack/tenks master: [WIP] Fix 'The `libvirt` module is not importable' error https://review.opendev.org/c/openstack/tenks/+/800199 | 10:54 |
| dtantsur | good afternoon ironic | 11:01 |
| opendevreview | Pierre Riteau proposed openstack/tenks master: [WIP] Fix 'The `libvirt` module is not importable' error https://review.opendev.org/c/openstack/tenks/+/800199 | 11:12 |
| arne_wiebalck | jssfr: doesn't Ironic send this (unless disabled)? | 11:24 |
| opendevreview | Pierre Riteau proposed openstack/tenks master: [WIP] Fix 'The `libvirt` module is not importable' error https://review.opendev.org/c/openstack/tenks/+/800199 | 11:28 |
| dtantsur | priteau: hi! have you figured out what causes ^^? | 11:31 |
| dtantsur | we have the same problem in bifrost (maybe somebody has already fixed it, dunno) | 11:32 |
| iurygregory | good afternoon dtantsur | 12:03 |
| opendevreview | Pierre Riteau proposed openstack/tenks master: Fix 'The `libvirt` module is not importable' error https://review.opendev.org/c/openstack/tenks/+/800199 | 12:12 |
| opendevreview | Dmitry Tantsur proposed openstack/bifrost master: Install libvirt-python from source instead of a wheel https://review.opendev.org/c/openstack/bifrost/+/800214 | 12:16 |
| dtantsur | trying the same workaround ^^^ | 12:16 |
| opendevreview | Dmitry Tantsur proposed openstack/bifrost master: Install libvirt-python from source instead of a wheel https://review.opendev.org/c/openstack/bifrost/+/800214 | 12:54 |
| dtantsur | iurygregory: if you have a second, this is upgrade-critical: https://review.opendev.org/c/openstack/ironic/+/799869 | 12:56 |
| iurygregory | dtantsur, sure | 12:57 |
| iurygregory | dtantsur, done | 12:59 |
| dtantsur | thanks! | 12:59 |
| iurygregory | np | 12:59 |
| iurygregory | going for lunch now (I totally forgot to order food early =X) | 13:00 |
| dtantsur | food shall not be forgotten! | 13:01 |
| opendevreview | Pierre Riteau proposed openstack/tenks master: Fix 'The `libvirt` module is not importable' error https://review.opendev.org/c/openstack/tenks/+/800199 | 13:04 |
| iurygregory | yeah, I forgot to order before 13:00 :D | 13:06 |
| cenne | happy lunch iurygregory | 13:11 |
| iurygregory | tks cenne :D and congrats on the merged patch :D | 13:12 |
| cenne | yess! thank you! | 13:12 |
| dtantsur | yeah, good job cenne | 13:13 |
| cenne | thanks to all of you. thanks for guiding me through it. and helping me when I was stuck. ^^ | 13:15 |
| cenne | merges feel good :) Triple green as opposed to single one on tests passing :p | 13:16 |
| TheJulia | good morning | 13:17 |
| dtantsur | morning TheJulia | 13:17 |
| iurygregory | good morning TheJulia | 13:17 |
| cenne | good morning TheJulia :) | 13:19 |
| dtantsur | fg | 13:34 |
| dtantsur | :( | 13:34 |
| cenne | fg/bg fg? | 13:40 |
| dtantsur | I keep forgetting which console I'm in :) | 13:40 |
| opendevreview | Merged openstack/tenks master: Fix 'The `libvirt` module is not importable' error https://review.opendev.org/c/openstack/tenks/+/800199 | 13:41 |
| dtantsur | iurygregory: it seems that https://review.opendev.org/c/openstack/bifrost/+/800214/ will pass the CI, could you check it as well please? | 13:41 |
| iurygregory | dtantsur, sure! | 13:43 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: Sanity check object version changes https://review.opendev.org/c/openstack/ironic/+/800252 | 13:43 |
| dtantsur | rloo: ^^^ | 13:43 |
| iurygregory | dtantsur, fyi I would have changed "name: install libvirt-python, gunicorn and lxml" but I'm ok to +2 if CI is green =) | 13:47 |
| dtantsur | ugh. lemme clean it up | 13:47 |
| iurygregory | up to you, CI seems happy in the centos8 jobs | 13:48 |
| dtantsur | I'll add a release note while I'm here | 13:48 |
| iurygregory | you can do a follow-up or update the patch, up to you =) | 13:48 |
| iurygregory | dtantsur, awesome tks | 13:48 |
| opendevreview | Dmitry Tantsur proposed openstack/bifrost master: Install libvirt-python from source instead of a wheel https://review.opendev.org/c/openstack/bifrost/+/800214 | 13:50 |
| dtantsur | iurygregory: here ^^^ | 13:50 |
| iurygregory | dtantsur, tks! | 13:50 |
| zaneb | arne_wiebalck: it's now fixed in RHEL btw. not sure if it made it into a release yet but I expect the fix should be in CentOS Stream | 13:51 |
| arne_wiebalck | zaneb: thanks! | 13:53 |
| zaneb | no worries | 13:55 |
| opendevreview | Dmitry Tantsur proposed openstack/ironic master: Drop support for pre-Victoria RPC/object versions https://review.opendev.org/c/openstack/ironic/+/800261 | 14:11 |
| dtantsur | rloo: also ^^ please if you have minute (it will need a rebase soon, but I'd like your opinion) | 14:11 |
| rloo | looking... | 14:11 |
| jssfr | arne_wiebalck, ironic does, I did not when using ipmitool manually :) | 14:14 |
| opendevreview | cenne proposed openstack/ironic master: [WIP] Add endpoint for setting node boot_mode on demand https://review.opendev.org/c/openstack/ironic/+/800084 | 14:16 |
| rloo | dtantsur: the only reason I can think of (now, anyway) keeping those is that it might be possible to do db migrations from an old eg pike to xena using xena code with some tweaks to the 'only support upgrades from previous named release'. I think that's why nova left (or used to leave, haven't looked recently) the conversions in the code. | 14:20 |
| rloo | dtantsur: am not sure where the community is going (if anywhere) wrt db migrations/upgrades beyond one release. | 14:21 |
| dtantsur | yeah, I guess the question is how long we should keep them | 14:22 |
| dtantsur | and how realistic it is to upgrade pike->xena without running dbsync/online migrations for all intermediate releases | 14:22 |
| dtantsur | cc TheJulia | 14:22 |
| rloo | true. i'm curious to know what the other openstack services do/support. | 14:23 |
| dtantsur | very few even support rolling upgrade :) | 14:23 |
| dtantsur | and I think people settled down on fast-forward, not skip-version upgrades? | 14:24 |
| rloo | i'm somewhat unopinionated, so let me know what you all think and I'll review more :) leaving the code there doesn't hurt but might cause confusion. | 14:24 |
| dtantsur | yeah, I'm not even sure it's going to work if you pin Xena to 9.2. too many changes since then. | 14:25 |
| rloo | i'm also wondering why we left the code in, if it was cuz i thought it was useful or cuz i stopped working on the upgrade stuff or didn't think about it. | 14:25 |
| * dtantsur votes for option #2 :) | 14:25 | |
| iurygregory | dtantsur, for the Refish Implementation details do you think http://paste.openstack.org/show/807343/ would be enough? | 14:25 |
| dtantsur | iurygregory: why require Context? | 14:26 |
| TheJulia | so, I recently had this whole discussion with someone regarding how frustrating the upgrade experience was | 14:27 |
| opendevreview | Verification of a change to openstack/ironic failed: Add missing BIOSSettings version mapping for older releases https://review.opendev.org/c/openstack/ironic/+/799869 | 14:27 |
| TheJulia | and how it really hails back to the days of when people did things manually | 14:27 |
| TheJulia | The newer experience/expectation really seems to be to "always just work, let the software sort it out database/content wise" | 14:27 |
| TheJulia | so instead of running commands, the upgrade is starting a newer version of ironic | 14:27 |
| TheJulia | and rolling forward | 14:27 |
| TheJulia | pike->xena should be fine, I think, we've not done much that requires transofrmations, we've used them to change things like iscsi->direct | 14:28 |
| dtantsur | this is kinda what metal3 does (except that it starts with a fresh database) | 14:28 |
| TheJulia | its not a difficult model reallly, and we already have most of the coding practices already, just not the actual "do the needful" | 14:29 |
| dtantsur | we could run dbsync on start-up and online_data_migrations after starting, but it's going to be interesting in the presence of several conductors | 14:29 |
| dtantsur | not sure it's related to the "remove pre-victoria" discussion though | 14:29 |
| TheJulia | of course, but we *need* to add logic to be cognizent of some of those things | 14:29 |
| dtantsur | we could use leader election for that. downside: some deployments don't want etcd/zookeeper | 14:30 |
| TheJulia | Well, there are a lot of deployments still running older releases and they may wish to jump. But nova recently went in the direction of "stop on ?train?" and everything will be okay | 14:30 |
| dtantsur | We don't support skip-version upgrades. Even if we have some code that may make it happen, all bets are off if you try. | 14:30 |
| TheJulia | dtantsur: eh, that is a technical detail and process detail | 14:30 |
| TheJulia | Well, people do it... *shrug* | 14:30 |
| dtantsur | I'm not aware of anyone | 14:31 |
| dtantsur | would be curious to hear any success stories about skip-version (NOT FFU) | 14:31 |
| TheJulia | I've heard some discussion, we've had the occasional person come in and imply they've done it | 14:31 |
| TheJulia | truthfully the thing that bites us is where we transform something with a migration | 14:32 |
| dtantsur | I'm against giving people a feel that you can upgrade N-2->N directly without anything actually backing (CI?) | 14:32 |
| opendevreview | Aija Jauntēva proposed openstack/sushy master: Fix OEM required attribute parsing https://review.opendev.org/c/openstack/sushy/+/800267 | 14:32 |
| TheJulia | I can agree with that | 14:33 |
| arne_wiebalck | Downstream, I would probably never allow an automatic migration of the DB, like a conductor with version N+1 connects to the DB and starts migrating it ... if that is what you're discussing. | 14:34 |
| TheJulia | There is a model of use, also with versioned objects which allows for a non-rewrite the db model | 14:34 |
| TheJulia | that is and the glue on startup is basically the parts of the interaction puzzle we miss | 14:35 |
| opendevreview | Aija Jauntēva proposed x/sushy-oem-idrac master: Add OEM Task extension https://review.opendev.org/c/x/sushy-oem-idrac/+/800269 | 14:37 |
| iurygregory | dtantsur, I added as required since the schema mentions that https://github.com/openstack/sushy/blob/master/sushy/resources/eventservice/eventdestination.py#L33 https://redfish.dmtf.org/schemas/v1/EventDestination.v1_0_0.json | 14:41 |
| dtantsur | iurygregory: oh, it was requiredOnCreate in 1.0.0? omg | 14:41 |
| iurygregory | yeah | 14:41 |
| dtantsur | what should a user put there though? | 14:41 |
| iurygregory | can be "" or any string | 14:41 |
| TheJulia | more event forwarding headaches? | 14:41 |
| dtantsur | iurygregory: then let's provide this default | 14:41 |
| iurygregory | we can ofc not ask and just say "Ironic" :D | 14:42 |
| dtantsur | empty string or "ironic" or whatever | 14:42 |
| dtantsur | yes, exactly\ | 14:42 |
| iurygregory | sure | 14:42 |
| dtantsur | let's make our API less painful than Redfish :D | 14:42 |
| iurygregory | yeah, give us a destination url and that's it... | 14:42 |
| TheJulia | indicating ironic may be useful if someone is trying to use multiple systems in a data center | 14:42 |
| dtantsur | if it's a free-form string - why not. we may even include the ironic version, just like a UserAgent :) | 14:43 |
| iurygregory | I don't want to test how is the free-form string... | 14:43 |
| iurygregory | I'm afraid the BMC will explode :D | 14:43 |
| dtantsur | possible :) | 14:44 |
| iurygregory | we can "add later" the support for context | 14:44 |
| dtantsur | iurygregory: what's it even for? | 14:46 |
| dtantsur | I mean, how is it used by the BMC? | 14:46 |
| iurygregory | This property shall contain a client supplied context that will remain with the connection through the connections lifetime | 14:46 |
| iurygregory | mraineri, do you have more info on how Context in EventDestination are used? | 14:47 |
| dtantsur | yeah, it's a confusing definition | 14:49 |
| dtantsur | what I care about is whether it's sent back with the notifications | 14:50 |
| dtantsur | if not, we don't need to expose it probably | 14:50 |
| mraineri | Context is used by the event receiver to help assist it with how to direct incoming events | 14:50 |
| mraineri | So, it's really a free-form string that a subscriber sets up to their liking, and the expectation is the service will simply echo the string back when publishing events | 14:51 |
| mraineri | The reason it's required to be specified on create is it's expected all events published outbound by Redfish will have a Context string | 14:52 |
| mraineri | Otherwise you get inconsistencies in the event payloads | 14:53 |
| iurygregory | gotcha | 14:53 |
| iurygregory | dtantsur, I think it would be useful to require this from the operator, because the system that will receive the Event Alerts can make different choices based on the context for example | 14:54 |
| dtantsur | what if it's not? | 14:54 |
| mraineri | If you have no use for doing any sort of top-level handling of events to direct them to different processes, then you could just statically use something like "Ironic" for your Context | 14:54 |
| dtantsur | this ^^ means the field is actually optional :) | 14:54 |
| dtantsur | iurygregory: I vote for an optional field. It's logically optional, and may not be supported in all drivers. | 14:55 |
| mraineri | Looking at the schema history now... It's actually NOT required on create anymore | 14:55 |
| mraineri | But it is required to be supported by the service | 14:56 |
| dtantsur | yeah, I also think it was required only in some Redfish versions | 14:56 |
| dtantsur | (which doesn't make anything easier tbh) | 14:56 |
| dtantsur | TheJulia: could you check https://review.opendev.org/c/openstack/bifrost/+/800214 please? fixes the bifrost CI. | 14:56 |
| iurygregory | optional defaults to "" or "Ironic" ? | 14:56 |
| iurygregory | and if the user wants to set it will be possible | 14:57 |
| dtantsur | iurygregory: I think it makes more sense to default to an empty string if it's allowed by the standard | 14:57 |
| dtantsur | yep | 14:57 |
| mraineri | Empty string is allowed | 14:57 |
| iurygregory | awesome | 14:57 |
| iurygregory | I will add this info in the spec | 14:57 |
| dtantsur | ++ | 14:59 |
| opendevreview | Verification of a change to openstack/ironic failed: Add missing BIOSSettings version mapping for older releases https://review.opendev.org/c/openstack/ironic/+/799869 | 15:02 |
| TheJulia | dtantsur: can after my meeting in a little bit | 15:11 |
| iurygregory | dtantsur, since we are adding extra we need to let user update right? asking because we are not considering PATCH in the API https://review.opendev.org/c/openstack/ironic-specs/+/785742/5/specs/approved/event-subscriptions.rst L27 and L161 | 15:20 |
| iurygregory | TheJulia, also if you have any insights on this ^ | 15:26 |
| iurygregory | can we just let the user set extra during create? :D | 15:26 |
| dtantsur | I'm fine with skipping patch for now | 15:26 |
| iurygregory | ack | 15:27 |
| TheJulia | Yeah, likely okay, but eventually someone will file a bug on not being able to update it for some reason | 15:27 |
| dtantsur | we can always follow-up :) I think allocations also started as immutable.. | 15:28 |
| iurygregory | yeah :D | 15:28 |
| dtantsur | I have to leave a bit earlier today, have a great weekend! | 15:31 |
| TheJulia | have a wondeful weekend! | 15:33 |
| arne_wiebalck | Bye everyone o/ | 15:45 |
| cenne | Bye arne_wiebalck, dtantsur | 15:47 |
| opendevreview | Iury Gregory Melo Ferreira proposed openstack/ironic-specs master: Event Subscription Spec https://review.opendev.org/c/openstack/ironic-specs/+/785742 | 15:47 |
| trandles | Question: Installing Bifrost from the master branch. In the "Validate API access and at least one conductor" task it's bombing out with a python traceback that it can't find the 'openstack' module. | 15:53 |
| trandles | well done trandles there's not actually a question in that...Fridays... | 15:54 |
| trandles | anyway, I've never seen a bifrost install crap out like that before | 15:55 |
| trandles | should I just activate the virtual env and install openstack using pip? | 15:55 |
| trandles | then re-run the bifrost install? | 15:55 |
| opendevreview | Merged openstack/tenks master: Test idempotence of deploy and teardown in CI https://review.opendev.org/c/openstack/tenks/+/660964 | 16:03 |
| TheJulia | trandles: so there is a known issue and a fix posted in CI | 16:08 |
| TheJulia | for at least a breaking issue, just haven't looked at it yet | 16:08 |
| trandles | TheJulia: where do I find that in CI? sorry, I've never accessed that before | 16:17 |
| TheJulia | review.opendev.org search for project:openstack/bifrost and it might be the patch posted that he asked for reviews on | 16:18 |
| trandles | ah, gotcha, thanks! | 16:18 |
| trandles | found dtantsur's patch but that's for libvirt-python not the openstack module | 16:28 |
| JayF | spuc? | 17:01 |
| opendevreview | cenne proposed openstack/ironic master: [WIP] Add api endpoint for setting boot_mode on live node https://review.opendev.org/c/openstack/ironic/+/800084 | 17:06 |
| cenne | TheJulia: For implementing setting boot mode from the api, I suppose I need to add another policy 'baremetal:node:set_boot_mode' | 17:06 |
| cenne | Can I skip the "deprecated_rule" because there never was? or do I need to back-add it? | 17:07 |
| cenne | ooh, spuc time. | 17:07 |
| JayF | No deprecated rule, it's a new feature thy should use RBAC imo | 17:10 |
| cenne | Okie. Thanks JayF | 17:12 |
| cenne | Is spuc still on? | 17:12 |
| JayF | we're in there | 17:16 |
| *** melwitt is now known as Guest322 | 17:31 | |
| TheJulia | ohhh | 17:44 |
| TheJulia | awww, too late | 17:45 |
| TheJulia | the fun of downstream meetings | 17:45 |
| TheJulia | cenne: you *can* skip the deprecated rule, or just use a legacy rule as fallback | 17:46 |
| TheJulia | just not everyone nor devstack by default uses for access | 17:46 |
| cenne | Yeah, spuc over unfortunately :( | 17:50 |
| cenne | You were missed :) | 17:51 |
| TheJulia | :( I feel bad, but sometimes I can't escape meetings on Friday | 17:51 |
| cenne | sorry :< | 17:57 |
| *** melwitt_ is now known as melwitt | 17:57 | |
| *** melwitt is now known as jgwentworth | 17:58 | |
| TheJulia | ahh, it is good old jgwentworth ;) | 18:03 |
| *** TheJulia is now known as needssleep | 18:04 | |
| jgwentworth | heh | 18:04 |
| cenne | heading out for the weekend folks. o/ | 18:32 |
| cenne | Have a nice weekend everyone! | 18:33 |
| cenne | Hope you get a very nice good sleep TheJulia | 18:36 |
| cenne | bye.~ | 18:37 |
| JayF | o/ | 18:41 |
| *** cenne is now known as cenn | 18:45 | |
| *** cenn is now known as cenne | 18:45 | |
| cenne | o/ JayF | 18:46 |
| *** cenne is now known as cenn | 18:46 | |
| opendevreview | Julia Kreger proposed openstack/ironic master: Add ipxe ramdisk kernel append param test https://review.opendev.org/c/openstack/ironic/+/800298 | 18:56 |
| opendevreview | Derek Higgins proposed openstack/ironic master: Allow reboot to HD following iso ramdisk deploy. https://review.opendev.org/c/openstack/ironic/+/800300 | 19:07 |
| opendevreview | Merged openstack/bifrost master: Install libvirt-python from source instead of a wheel https://review.opendev.org/c/openstack/bifrost/+/800214 | 19:25 |
Generated by irclog2html.py 2.17.2 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!