Details
-
Task
-
Resolution: Done
-
Undefined
-
None
-
None
-
None
-
False
-
None
-
False
-
Testable
Description
Originally, the ELN buildroot sync did not have --delete in the command, and it started using alot of disk space.
We now have --delete with every sync, each hour, and it's causing it's own problems.
If there is an ELN update to a package that is needed for a build, there is now a 5 to 15 minute timeframe that all builds on both CS10 and RHEL10 will fail. And that is assuming that the CS and RHEL buildroots start their buildroot regeneration immediately after the external repo sync happens.
When we are doing mass builds, that is affecting alot of builds.
My proposal is this. Have two syncing scripts. One with --delete, one without. During the hourly syncs, use the script without delete. Then once a week, during what should be a slow time (Saturday 00:00 UTC), run the script with --delete.
If it turns out it is still taking too much space, then try daily at a slow time (00:00 UTC).