[HTCondor-users] Limit total STARTER log size across many slots?

["Fischer, Max (SCC)" <max.fischer@xxxxxxx>]

Hi all,

Iâve just dialled down `MAX_STARTER_LOG` since on our larger worker nodes (192-256 cores) logs were taking up unexpectedly much space. The reason for this is clear to us: For each core we get a `StarterLog.slot1_\d+` plus `StarterLog.slot1_\d+.old` - i.e. a worst case of up to 500 times  `MAX_STARTER_LOG`!
However, just reducing `MAX_STARTER_LOG` isnât quite ideal. Due to differently sized jobs, `StarterLog.slot1_1` sees much more action than, say, `StarterLog.slot1_192` and the two rotate at wildly different rates (currently days versus months).

Are there any HTCondor knobs for limiting the total STARTER logs size while achieving a more uniform log rotation?

Cheers,
Max