Trying to backup to a datastore stored on a remote Samba share (Hetzner Storage Box) causes the server to become almost completely unresponsive after some amount of data has been uploaded to it. On the remote end, upload usually times out when this happens. Before this happens, data upload appears to work fine. (I did try monitoring network activity on the PBS system, found nothing unusual).
Main symptom is kswapd0 kernel daemon using all the CPU time. After stopping the backup and waiting for a couple minutes (or killing PBS daemon), kswapd0 activity stops.
Sometimes, the server not just hangs, but also reboots itself (likely due to a kernel panic, but I haven't been able to capture this).
Network connection to the storage box is stable there are no issues with it (apart from speed varying a lot, but AFAIK, this is to be expected). Internet connection is gigabit fiber.
PBS runs on a KVM VM, specs: 4 cores of an i5-8400, 8 GB memory, 4 TB local storage (for onsite backups). Swap is NOT enabled. Memory usage seems fine (according to "free" command) even when symptoms occur.
Client is PVE running on another machine, encryption is enabled (if that matters). Using PBS server's local disk for onsite backups works fine.
What could be the cause ?
Main symptom is kswapd0 kernel daemon using all the CPU time. After stopping the backup and waiting for a couple minutes (or killing PBS daemon), kswapd0 activity stops.
Sometimes, the server not just hangs, but also reboots itself (likely due to a kernel panic, but I haven't been able to capture this).
Network connection to the storage box is stable there are no issues with it (apart from speed varying a lot, but AFAIK, this is to be expected). Internet connection is gigabit fiber.
PBS runs on a KVM VM, specs: 4 cores of an i5-8400, 8 GB memory, 4 TB local storage (for onsite backups). Swap is NOT enabled. Memory usage seems fine (according to "free" command) even when symptoms occur.
Client is PVE running on another machine, encryption is enabled (if that matters). Using PBS server's local disk for onsite backups works fine.
What could be the cause ?
Last edited: