The artifact cleanup task does so much work (allocating a lot of objects and maybe also run a lot in parallel).

During the execution of the task the request queue gets much slower and a lot of builds remain in the CHECKING_BUILD_CONDITION status.

Also we had a lot of configurations which had the red error field on top stating "GC overhead limit exceeded".

Can you please look into this.
Yesterday we had to do a emergency restart of QB server because of that.
Today it was again but luckily it recovered and works normal again.