| *** rlandy_ is now known as rlandy | 11:27 | |
| tkajinam | o/ I wonder if it's possible to hold a vm in CI for some debugging ? | 12:06 |
|---|---|---|
| tkajinam | would like to get into the node of heat-functional job (which is being triggered by https://review.opendev.org/c/openstack/heat/+/937617 ) | 12:07 |
| frickler | tkajinam: sure, give me a minute. you can also let me know your ssh public key already | 12:13 |
| frickler | ok, should be in time to hold the current run assuming it will fail | 12:15 |
| tkajinam | frickler, the public key is https://github.com/kajinamit.keys | 12:16 |
| tkajinam | and thanks ! :-) | 12:17 |
| tkajinam | can it be kept for a few days ? I'll check details likely tomorrow and will also ask others to have a look | 12:18 |
| frickler | tkajinam: sure, just let us know once you are done. root@65.17.193.167 - note the build is still running, but I'm assuming it will fail based on the history, so I added you there already. just be aware that you might get weird results if you touch anything before it fails | 12:21 |
| tkajinam | frickler, thanks and I understand that. I'll touch services once the current run completes (and fails, I assume) | 12:22 |
| frickler | if it passes, the node will be deleted and we'll have to try again with another run | 12:23 |
| tkajinam | ah, ok. If that passes there may be no problems we have to dig into :-P | 12:23 |
| tkajinam | frickler, I just succeeded to login to the node. will monitor the run. | 12:24 |
| frickler | tkajinam: don't worry, if that problem is solved, I've got the next problem waiting for you in the queue, named eventlet bump ;-D | 12:27 |
| tkajinam | :-P | 12:39 |
| tkajinam | the job failed \o/ | 13:39 |
| * tkajinam signs | 13:39 | |
| * tkajinam sighs | 13:39 | |
| frickler | yay ;) | 13:40 |
| stephenfin | fungi: clarkb: We've hit rate limits in OSC jobs recently. Two questions: (a) has anything changed there to be causing this, and (b) should we be running this in post rather than gate? https://review.opendev.org/c/openstack/python-openstackclient/+/885517 | 15:45 |
| fungi | stephenfin: dockerhub rate limits specifically? | 15:50 |
| stephenfin | sorry, yes | 15:50 |
| fungi | dockerhub changed their rate limit quotas recently to be far more restrictive. most projects are working out options for moving off dockerhub as much as possible, and in some cases mirroring their dependencies somewhere else too | 15:51 |
| stephenfin | ack, so not something that's changed here. I'll look at migrating to periodic and adding a quay.io job | 15:51 |
| stephenfin | here being opendev | 15:51 |
| fungi | we also discovered yesterday that dockerhub counts all addresses within an ipv6 /64 network as a single host for quota purposes | 15:52 |
| fungi | so for cloud providers who assign all server instances public addresses out of a single /64 network, it's functionally as if they're all being funneled through a single-address ipv4 overload nat rate limit wise | 15:53 |
| fungi | those two factors combined have made using dockerhub in ci jobs nigh unworkable | 15:53 |
| stephenfin | ack. Dumb question then: how do I get credentials for quay.io? I assume I should be publishing to a opendev or openstack namespace | 15:56 |
| fungi | stephenfin: i think each team has their own namespace in quay and sets up an account that they put credentials for in a zuul secret | 15:58 |
| clarkb | right we don't moderate third party account credentials like that | 15:59 |
| clarkb | zuul has an account and opendev has one and kolla has one and so on | 15:59 |
| clarkb | each being their own responsibility. | 15:59 |
| fungi | e.g. https://quay.io/repository/openstack.kolla/fluentd | 16:00 |
| stephenfin | Hmm. I wonder if we should simply not publish images... | 16:01 |
| fungi | we're also, separately, working on solutions to reduce overall request volume to outside container registries in jobs, like pre-caching containers in ephemeral buildset registries so multiple jobs in the same buildset can reuse them | 16:02 |
| clarkb | I guess we do haev a central pypi account but that is mostly for historical reasons when taht was easier with jenkins | 16:02 |
| clarkb | fungi: well to be clear opendev already provides the functionality to do that via the buildset registry. It is up to jobs to take advantage of it | 16:02 |
| clarkb | I don't know that there is much more for us to do with the precaching idea. | 16:03 |
| fungi | other than have examples to point to, yeah | 16:03 |
| *** Guest2639 is now known as diablo_rojo_phone | 16:22 | |
| gmann | haleyb|out: things working here now, can you remove the WIP from this and its depends-on? https://review.opendev.org/c/openstack/charm-mysql-innodb-cluster/+/937195 | 19:24 |
| haleyb|out | gmann: thanks for the review, but i don't think i defined the openstack-python3-charm-jobs-ubuntu correctly, i actually wanted it to trigger the distro-specific templates below it. I think i should just remove it | 20:27 |
| gmann | haleyb|out: ok, that also work and use distro specific template in distro specific branches | 20:28 |
| haleyb|out | gmann: right, since the jammy one should have py38 and py310 voting, etc, i'm pushing an update soon | 20:28 |
| gmann | haleyb|out: +1, good point | 20:29 |
| haleyb|out | i think that was actually broken now that i'm looking | 20:29 |
| gmann | haleyb|out: I will suggest to keep it same way which is n-v in current template and later in separate change fix it and run as voting | 20:52 |
| haleyb|out | gmann: done | 21:03 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!