| gmann | kopecmartin: dansmith: as master grenade change is merged, these are backports to stable https://review.opendev.org/q/I05fcdb5045a33997bd1a4c340a16458d88399f5f+status:open | 00:12 |
|---|---|---|
| opendevreview | Merged openstack/grenade stable/2023.1: Fix tempest venv constraints for target branch tempest run https://review.opendev.org/c/openstack/grenade/+/879119 | 09:05 |
| TheJulia | dansmith: I *think* I've figured out one of the two cases we've encountered, tl;dr cirros's networking might not be online before we give up trying to ping in the entire sequence of events when it is a deploy via ironic, the case we were looking at logs going "what in the world" yesterday, remains a mystery at the moment | 16:28 |
| dansmith | that's a very common manifestation, and I'm not sure it's really cirros' fault | 16:28 |
| dansmith | but we have lots of failures where tempest spins up a guest with cirros, and then can just never ping/ssh it | 16:29 |
| dansmith | I think it's more actual network plumbing because usually cirros knows what its networking should be because of the metadata server | 16:29 |
| dansmith | but, I didn't see where that was coming into the investigation yesterday.. is that just another common failure you see? | 16:30 |
| opendevreview | Julia Kreger proposed openstack/grenade master: [ironic] Add an instance wait setting for nova resources https://review.opendev.org/c/openstack/grenade/+/879674 | 16:57 |
| JayF | dansmith: we basically have two known issues in grenade right now: the flavor thing y'all were chasing yesterday, and a thing where a vip doesn't ping. The node that got held was the 'vip doesn't ping' failure case. | 16:59 |
| dansmith | ack | 16:59 |
| JayF | which 879674 gives us the knobs to fix | 16:59 |
| dansmith | yeah, I'm dubious about that needing to be longer for a VM, but perhaps the ironic case for an override is reasonable | 17:00 |
| frickler | fwiw cirros 0.6.1 ignores all network metadata, it just tries to dhcp everywhere | 17:04 |
| frickler | it also should get made better at waiting for networking to be available https://github.com/cirros-dev/cirros/issues/97 | 17:05 |
| dansmith | uh, really? that's news to me (obviously).. we tried a switch to newer cirros and failed because of network issues, perhaps that's why? | 17:07 |
| dansmith | I assume that's not the issue here though | 17:08 |
| frickler | depends on what networking issues, but might be an option, yes. who is "we" in your context? | 17:08 |
| dansmith | I think bauzas floated a patch | 17:09 |
| dansmith | I meant not an issue here because the existing jobs are using 0.5.x AFAIK | 17:09 |
| opendevreview | Julia Kreger proposed openstack/grenade master: [ironic] Add an instance wait setting for nova resources https://review.opendev.org/c/openstack/grenade/+/879674 | 17:56 |
| opendevreview | James Parker proposed openstack/whitebox-tempest-plugin master: Add vdpa base movement tests https://review.opendev.org/c/openstack/whitebox-tempest-plugin/+/871510 | 21:11 |
| opendevreview | Merged openstack/grenade stable/zed: Fix tempest venv constraints for target branch tempest run https://review.opendev.org/c/openstack/grenade/+/879123 | 22:20 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!