What feature do you want to see added?
I just went down the rabbit hole figuring out why our Jenkins UI was slow at times, which then might us notice regular high disk utilization on our Grafana monitoring instance.
This is the Grafana reporting on our Linux VM hosting the Jenkins installation:
As you can see after a quiet period of 15 mins there's a highly active period of around 25 mins where there's lots of disk I/O (mostly reads).
That lead me to inspecting the Jenkins' JVM threads where I stumbled upon:
"Simple disk usage computation [#1]" Id=332 Group=main RUNNABLE
at java.base@17.0.17/sun.nio.fs.UnixNativeDispatcher.readdir(Native Method)
at java.base@17.0.17/sun.nio.fs.UnixDirectoryStream$UnixDirectoryIterator.readNextEntry(Unknown Source)
at java.base@17.0.17/sun.nio.fs.UnixDirectoryStream$UnixDirectoryIterator.hasNext(Unknown Source)
- locked sun.nio.fs.UnixDirectoryStream$UnixDirectoryIterator@79673db3
at java.base@17.0.17/java.nio.file.FileTreeWalker.next(Unknown Source)
at java.base@17.0.17/java.nio.file.Files.walkFileTree(Unknown Source)
and then https://github.com/jenkinsci/cloudbees-disk-usage-simple-plugin/blob/master/src/main/java/com/cloudbees/simplediskusage/QuickDiskUsagePlugin.java#L59 (which in fact is that 15 mins quiet period I was experiencing in the monitoring).
Arguably our installation is very large (Jenkins' home is multiple 100Gs), but still this plugin should probably take some precautions to not cause such high disk I/O.
Some random idea to mitigate this at least:
Maybe add some warning popup for admins in Jenkins whenever that scan jobs takes more than, say, 5 minutes? So you'd at least notice that something (bad) is going on. It's difficult to even notice that without the proper monitoring.
Upstream changes
No response
Are you interested in contributing this feature?
No response
What feature do you want to see added?
I just went down the rabbit hole figuring out why our Jenkins UI was slow at times, which then might us notice regular high disk utilization on our Grafana monitoring instance.
This is the Grafana reporting on our Linux VM hosting the Jenkins installation:
As you can see after a quiet period of 15 mins there's a highly active period of around 25 mins where there's lots of disk I/O (mostly reads).
That lead me to inspecting the Jenkins' JVM threads where I stumbled upon:
and then https://github.com/jenkinsci/cloudbees-disk-usage-simple-plugin/blob/master/src/main/java/com/cloudbees/simplediskusage/QuickDiskUsagePlugin.java#L59 (which in fact is that 15 mins quiet period I was experiencing in the monitoring).
Arguably our installation is very large (Jenkins' home is multiple 100Gs), but still this plugin should probably take some precautions to not cause such high disk I/O.
Some random idea to mitigate this at least:
Maybe add some warning popup for admins in Jenkins whenever that scan jobs takes more than, say, 5 minutes? So you'd at least notice that something (bad) is going on. It's difficult to even notice that without the proper monitoring.
Upstream changes
No response
Are you interested in contributing this feature?
No response