#10721 [modularity][arch-specific] KOJI armv7hl builders infinite restarting
Closed: Fixed with Explanation by kevin. Opened by mschorm.

In the past few days, I've submitted a number of builds of MariaDB and MariaDB modules for 4 module streams.

From Monday, I keep seeing the armv7hl behaviour where the builder restart the task again and again without any error message.

Until yesterday the vast majority of cases was F34-specific, so I just ignored that release, in a belief it's not worth the trouble since F34 will be EOL in a few days - the updates won't even go through BODHI fast enough.

Today, while building the 10.8 stream of the mariadb module, I've been hit by this issue on F>34 too.

https://release-engineering.github.io/mbs-ui/module/14533/components
https://koji.fedoraproject.org/koji/taskinfo?taskID=87480912
https://release-engineering.github.io/mbs-ui/module/14534/components
https://koji.fedoraproject.org/koji/taskinfo?taskID=87476085
https://release-engineering.github.io/mbs-ui/module/14535/components
https://koji.fedoraproject.org/koji/taskinfo?taskID=87473764

Since technically, I am not the author of those tasks - MBS is - I can't cancel them one by one.
I can only cancel them all by fedpkg module-build-cancel and re-submit them, but it would be probably better to keep running just one F release at a time to ease the armv7hl builder burden.

Screenshot_from_2022-05-25_20-34-52.png


ok. It seems the 32bit arm builders had dnf anoyingly install kernel-core on them. ;( So, only a total of 3gb memory was available on them. ;(

I've removed that and rebooted them all into the lpae kernels as they are supposed to be using. ;(

So, it should be back to normal now. If you note this still happening, please re-open, but it should be fixed.

Existing builds should have restarted and now be running again on the lpae booted builders.

Metadata Update from @kevin:
- Issue close_status updated to: Fixed with Explanation
- Issue status updated to: Closed (was: Open)

Metadata