Closed tuminoid closed 3 months ago
/override metal3-centos-e2e-integration-test-main metal3-ubuntu-e2e-integration-test-main GH action.
@tuminoid: Overrode contexts on behalf of tuminoid: metal3-centos-e2e-integration-test-main, metal3-ubuntu-e2e-integration-test-main
/cc @kashifest @Rozzii
/cherry-pick release-24.1
@tuminoid: once the present PR merges, I will cherry-pick it on top of release-24.1 in a new PR and assign it to you.
/cherry-pick release-24.0
@tuminoid: once the present PR merges, I will cherry-pick it on top of release-24.0 in a new PR and assign it to you.
/cherry-pick release-23.1
@tuminoid: once the present PR merges, I will cherry-pick it on top of release-23.1 in a new PR and assign it to you.
/cc @NymanRobin
/lgtm
The change is good. However, please note that there have been some issues with VM scheduling for the jobs. For instance, this job took 15 hours when scheduling time is included: Build #62.
Maybe these jobs could be moved from Jenkins to github action runners in the future to improve reliability
WOW that is not expected IMO, anyways yeah I would guess container image building could be a github action, I think that was considered originally too.
The change is good. However, please note that there have been some issues with VM scheduling for the jobs. For instance, this job took 15 hours when scheduling time is included: Build #62.
Maybe these jobs could be moved from Jenkins to github action runners in the future to improve reliability
Yes agreed. We can see that after scheduling, the build itself took almost 21minutes (~1250 sec), which would've timeouted as well, so this fix is needed. It would then leave 30mins for scheduling as it is counted as well. Similarly for sushy/vmbc, the safety allows ~10 minutes of scheduling. We should be able to accomplish that in Jenkins or with runners. Jenkins scheduling we can fix ourselves, but runners would be at mercy of CNCF and other runner users.
edit: Correcting myself: for runners scheduling is not an issue, since if the workflow isn't scheduled, the time isn't running. Doh :)
/cc @dtantsur @elfosardo
/approve
[APPROVALNOTIFIER] This PR is APPROVED
This pull-request has been approved by: dtantsur, kashifest, NymanRobin
The full list of commands accepted by this bot can be found here.
The pull request process is described here
@tuminoid: new pull request created: #520
@tuminoid: new pull request created: #521
@tuminoid: new pull request created: #522
Ironic-image build locally, with a fast disk, fast CPU, and fast net takes 500-600s, depending on pre-pulls, repository speeds etc. New cloud has none of those reliably, so it times out at 1000s far too often. Also scheduling time is considered when images are built in Jenkins, and often scheduling takes a long time. Thanks to #505 we get notifications of these failures now.
I will cherry-pick this to release branches as well.