Recent content by Lephisto

I am experiencing the same, did you come to any conclusion?

Hi, it would be really nice to have an option that pauses verify Jobs while Tape Backups are running. Especially on PBS System with spinning rust (yes, they exist, and they are probably the vast majority of PBS installations out there) verify jobs cause the tape backup rate to drop and a...

I just noticed that for spinners it might be cruciial to also set: echo 12000 > /sys/module/zfs/parameters/zfs_arc_min_prefetch_ms echo 10000 > /sys/module/zfs/parameters/zfs_arc_min_prescient_prefetch_ms This way I somehow reliably max out the drives speed at 300MB/s.

So an Update on that Issue. It seems that the "Eject Media" Option in Fact seems to fix the stability Issues with the Quantum Superloader 3 (In my case with a LTO9 drive). The Syslog is now clean from SCSI Errors, and Backups run reliably. This is _not_ the "eject-before-unload" Option which...

It is not totally clear yet, I replaced all possible Hardware several times now. From what it seems, checking the Option "Eject Media" at the Backup job seems to do something. At least since about a week the Backup and Library is running stable.

Ideally you would sustain the Performance of what the drives can do, because you don't want them to throttle down every few seconds. That's 300MB/s for a LTO9-HH drive, I don't know what LTO-6 can do.

Done that.

So, as for the moment, the patch seems to work fine, even with verification jobs running in parallel it maxes out over large stretches of the time. #worksforme

You know that this is not the reality. No one uses flash as the only Backup Medium, just because PBS suffers from bad technical design decisions. Even today it's just too expensive.

LTO9HH (half-height) drives actually max out at 300MB/s, that's reasonable.

Okay, I have a more streamlined Hardare setup I guess.. AMD Epyc 7313 16-Core / 128GB ECC / Broadcom/LSI Tri-Band Controller / Exos 20TB Drives / 4-way mirror for special device on NVME (2x Koixia, 2x Samsung) and the LTO9HH drive in a Superloader 3. Workers set to 16, and even if no other...

Well for me it also doesn't totally sustain 300MB/s all the time, but it's okay'ish. @pibber what cpu and hdd/controller type do you run?

I tried the latest patch (https://forum.proxmox.com/threads/problems-compiling-pbs-v3-3-2.161721/) against 3.3.4. Mixed results. Somehow i don't seem to be able to sustain speeds. It's the only task running. Is there something significantly different? //edit ping @dcsapak

I have it running now. I removed the old tuning params from the datastore.cfg. How are the worker-threads defined now?

@fabian thanks for fixing!