S3 Garbage Collection Seems Stuck, PBS 4.1.4

[Mike_Scholes]

Hi all.

I'm running a garbage collection on my S3 backup after a prune job. I'm anticipated several TBs of data to be removed. The process went through the marking stage with all the usual "marked x% (x of x index files)" listed in a reasonable time. Then it seems to have stopped on "Start GC phase2 (sweep unused chunks)" and not moved on after 3.5 hours. On my home system I do get the list showing "processed x% (x chunks)" Anything I can check to see something is happening. Nothing I can see in syslog, very little activity, CPU, IOPS, Network etc.

 

[Mike_Scholes]

Quick update, looking at previous GC jobs phase 2 does appear to be a big operation with a single step that does take some time. It then goes onto phase 3 with percentages. As this is a lot of data it looks like phase 2 will be a long process. I'll post back when/if something happens.

 

[Robert Obkircher]

Hi,

This step does not yet output any progress logs, so it probably just needs more time.

You could check if the number of marker files in the local cache changes:

find <datastore-path>/.chunks -type f -print | wc -l

More detailed logging is already being worked on [1].

[1] https://lore.proxmox.com/pbs-devel/20260121104537.495434-1-c.ebner@proxmox.com/T/

 

[Mike_Scholes]

The process did eventually complete so we are good on this cheers. That extended logging will be useful, knowing things are still progressing is way better that nothing.