@davidjericho 200 threads for the conversion and the draining seems a bit to high.. in general we run a slow pace balancing with few transfers per node (between 2 and 50), how about the rate, what is your setting for balancer.node.rate?
@davidjericho which threshold do you set for the balancing? (balancer.threshold)
If you have reference to a shadow filesystem, which does not exist anymore, you can still dump it with the command (if you remember the fsid). If you have references to filesystems which are not configured anumore, get you a shadow filesystem entry hint by fsck.
@apeters1971 I've not had problems or errors with the FST crashing since maybe 0.3.166 or so, no reported errors for the balancing that I've found
I realise I could well have some very wacky space settings, I have to admit despite reading the documentation and looking into the code a few times, I'm not totally understanding what each space balancing/draining setting does and how they interplay