The Eclipse Jenkins for Eclipse Kura seems stuck. We have more than 20 jobs in the waiting list to be built. The usual 2 executors that should pick the jobs in the queue list are not available and even not displayed
Steps to reproduce
Access the Eclipse Jenkins for Eclipse Kura https://ci.eclipse.org/kura/ on the left part you'll see the queue for build filled with jobs and in the lower left part no running executors
What is the current bug behavior?
No build is performed. The project has multiple pull request waiting
What is the expected correct behavior?
Build should be pipelined and executed
Relevant logs and/or screenshots
Priority
Urgent
High
Medium
Low
Severity
Blocker
Major
Normal
Low
Impact
The Eclipse Kura project is currently blocked as no pull request can be built. Pull requests, as a result cannot be merged.
I'm sorry, I have to reopen it again.
Now the build works but we are running on only one executor instead of the usual 2 and it is making the whole process extremely slow and tedious.
Can you have please a check to see if an action needs to be taken?
On a (probably) related note: we noticed that there seems to be a periodic build active for our project. Every day it builds all the open PRs for our repo which is kinda wasteful and (looks like) increases the chances of Jenkins tripping. Can we do something for that?
I've restarted the Kura Jenkins instance and builds are running again.
On a (probably) related note: we noticed that there seems to be a periodic build active for our project. Every day it builds all the open PRs for our repo which is kinda wasteful and (looks like) increases the chances of Jenkins tripping. Can we do something for that?
In your multibranch configuration (https://ci.eclipse.org/kura/job/multibranch/configure), the option "Periodically if not otherwise run -> Interval -> 1 day" is set. So every day, all Branches and PRs are indexed. Builds should only be triggered if something has changed.
Since there are 30 open PRs it's hard to tell if builds are triggered more often than necessary. If you can pinpoint such a case, please let us know.
I've restarted the Kura Jenkins instance and builds are running again.
Thank you for the quick feedback
Since there are 30 open PRs it's hard to tell if builds are triggered more often than necessary. If you can pinpoint such a case, please let us know.
This is a screenshot of the queue taken yesterday
PR 2534 is 4 years old at this point and nothing changed for that but it's still in the queue. Apart from this, the behaviour is what we expect from our settings.
Since this issue is reappearing regularly on our CI (it is a weekly recurrence lately) and it is fairly painful to deal with, do you have any suggestion on how to avoid this? Is there anything we can do?
Checking pull request #2534 ‘Jenkinsfile’ found Met criteriaChanges detected: PR-2534 (40b6d1f86e0c32d2d7a017b3ede22520e92e1727+4ccc70c140d39fa27cb18035af3feb452b8d7631 (56c383c8b59ce9b22baf35931e7cc2f3a07b922a) → 40b6d1f86e0c32d2d7a017b3ede22520e92e1727+d52adb481d8342375361e0d38b6d413622a26579 (f8d6d7aca4eda6da8f7abdbc0c5b9dedcf71508a))Scheduled build for branch: PR-2534
There is a known issue with the Jenkins Kubernetes plugin that can lead to stuck builds, like you regularly experience at the moment. Usually this occurs only occasionally. Your CI instance seems to be affected a lot more often, though.
The issue described in #2816 (closed) lead to failed builds.
=> this is hopefully fixed now
My general recommendation for putting less pressure on the Jenkins instance is to reduce the number of open PRs. Especially old or outdated PRs.
We're experiencing the usual issue with the excecutors... but hey, it seems that the situation is improving. At least I'm not pestering you every other week
As you might notice we applied your suggestion and got rid of all the old open PRs reducing the load on the Jenkins instance.
We're experiencing the usual issue with the excecutors... but hey, it seems that the situation is improving. At least I'm not pestering you every other week
I was starting to miss our conversations.
As you might notice we applied your suggestion and got rid of all the old open PRs reducing the load on the Jenkins instance.
The Kura Jenkins instance has been restarted. Build is running again.
@mmaiero As a quick workaround I did the usual thing - restarted Jenkins. I don't have long term solution but will also add this problem to my watchlist.
Meantime, would it be possible to reduce build history size for all jobs to something like 5? we would like to transfer this instance to the new backend storage.
Once the transfer is complete we can restore those values.
Meantime, would it be possible to reduce build history size for all jobs to something like 5? we would like to transfer this instance to the new backend storage.
We updated our builds according to what you suggested.
Hello @pstankie@heurtemattes,
we're currently experiencing some issues with our Jenkins instance. Only one executor is currently available and builds are piling up.
Hello @pstankie@heurtemattes,
we're experiencing a severely degraded Jenkins instance currently. The UI is barely responsive and build fail to trigger. Is there something you can do on your end?