I can't say for sure, but my thoughts:
- It probably has nothing to do with pacemaker, cluster and drbd
- SQL databases use garbage collector, which could do one big transaction
on restart and freed some space
- Filesystem itself may be using some sort of garbage collector, especially
on remounting
>>> Eric Robinson schrieb am 04.10.2016 um 09:24 in
Nachricht
> The filesystem on my corosync+pacemaker cluster is 1TiB in size and was 95%
> full, with only 54GB available. Drbd was UpToDate/UpToDate.
>
> I restarted Pacemaker, and after that my filesystem now shows 49% full with
> 300GB+ f