| tonyb | This may be better on the zuul matrix. I have (well I'm creating) a multinode job and something seems to be serializing the hosts such that the run playbook for my job only starts on the second host after it's completed on the first. The pre playbooks seem to be running at the same time. | 00:26 |
|---|---|---|
| tonyb | I'm guessing there is a semaphore or similar causing that to happen. In my case I really want the 2 hosts running at the same time. How can I do that? | 00:26 |
| tonyb | For Reference: https://1d54657e8219699713ac-7bc7b92e78ec9846ffad73c4c2170bad.ssl.cf1.rackcdn.com/906022/17/check/requirements-generate-constraints-multinode/0e79106/job-output.txt | 00:27 |
| clarkb | tonyb: I think by default what you want should be done. For example in this task you can see interleaving TASK [configure-mirrors : Update apt cache] | 00:38 |
| clarkb | I do agree that later tasks seem to be running in serial though | 00:39 |
| tonyb | clarkb: I guess that's the default ansible behavior now that I think about it. | 00:43 |
| clarkb | I'm confused I'm looking at the job output and it doesn't seem to map to the playbooks in the change? | 00:43 |
| tonyb | clarkb: In what way? | 00:43 |
| clarkb | I think I see why. It is because you are including role(s) not in the change diff | 00:44 |
| clarkb | er tasks not roles. That explains my confusion | 00:44 |
| tonyb | Oh the bindep role ? | 00:45 |
| clarkb | no generating constraints | 00:47 |
| clarkb | https://groups.google.com/g/ansible-project/c/HcrRVFvA-24/m/znovAJ6oCAAJ I think this is realted and it is the include_tasks that serializes | 00:47 |
| clarkb | I don't fully understand why that is happening but it seems people haev observed this behavior and it is expected? | 00:48 |
| tonyb | I'm going to try adding a 'strategy: free' as that explicitly says slow hosts wont slow down faster hosts. *within* a host it will serialise but I'm okay with that. | 00:49 |
| tonyb | If I really wanted it to go "faster" I'll just use more hosts! | 00:50 |
| tonyb | In other news, I deleted a few inmotion nodes and a new one immediately got created and is active \o/ | 00:54 |
| clarkb | yay. Did you have to do anything special to delete the nodes? | 00:56 |
| tonyb | Nope just openstack ... server delete | 00:57 |
| clarkb | hrm | 00:57 |
| clarkb | did you do that with the admin credentials? | 00:57 |
| clarkb | maybe that is the difference | 00:57 |
| tonyb | Nope. | 00:57 |
| clarkb | weird I tried deleting a host and it said that uuid didn't exist | 00:58 |
| tonyb | I tried to do someing in the API and it said you need to be an admin | 00:58 |
| clarkb | maybe its a subset with issues | 00:58 |
| tonyb | Yeah. the top one in the list says that | 00:58 |
| clarkb | ha I got lucky picking the first one | 00:58 |
| tonyb | so far for the nodes stuck in "BUILD" it's 50:50 | 00:59 |
| tonyb | Okay I've removed what I can, now we're at the point where there are instances in the 'server list' that are only partially there | 01:08 |
| tonyb | I think based on some spot-checking those are the same nodes in nodepool listed as 'deleteing' | 01:09 |
| tonyb | Going back. Adding `strategy: free` has done what I wanted. The output is much harder to read but things are running at the same time #huzzah | 01:31 |
| opendevreview | Merged openstack/diskimage-builder master: Replace OpenSUSE Leap 15.3 to OpenSUSE Leap 15.5 https://review.opendev.org/c/openstack/diskimage-builder/+/904821 | 01:37 |
| opendevreview | Merged openstack/diskimage-builder master: python3.12: "fix" unittests https://review.opendev.org/c/openstack/diskimage-builder/+/903848 | 01:42 |
| opendevreview | Jeremy Stanley proposed opendev/system-config master: DNM: Fail keycloak testing for an autohold https://review.opendev.org/c/opendev/system-config/+/906600 | 02:16 |
| opendevreview | Merged openstack/diskimage-builder master: Fix small syntax issue https://review.opendev.org/c/openstack/diskimage-builder/+/903980 | 02:31 |
| opendevreview | Merged openstack/diskimage-builder master: Fix indentation of output formats https://review.opendev.org/c/openstack/diskimage-builder/+/896610 | 02:31 |
| opendevreview | Jeremy Stanley proposed opendev/system-config master: Switch from legacy to new style keycloak container https://review.opendev.org/c/opendev/system-config/+/905469 | 19:17 |
| opendevreview | Jeremy Stanley proposed opendev/system-config master: DNM: Fail keycloak testing for an autohold https://review.opendev.org/c/opendev/system-config/+/906600 | 19:17 |
| opendevreview | Jeremy Stanley proposed opendev/system-config master: Switch from legacy to new style keycloak container https://review.opendev.org/c/opendev/system-config/+/905469 | 19:44 |
| opendevreview | Jeremy Stanley proposed opendev/system-config master: DNM: Fail keycloak testing for an autohold https://review.opendev.org/c/opendev/system-config/+/906600 | 19:44 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!