Thank you every visitors and for those who did not had the chance to visit fosdem check this report:
https://xcp-ng.org/blog/2026/02/19/fosdem-2026-follow-up/
[image: vates-xcp-ng-at-fosdem-2026-1.webp]
One question, I promised to forward to the forum, how many VM are you running on XCP-ng ?
Dozen, hundreds, thousands or more ?
@comdirect
Use this command: (replace sda in the command below with the relevant device)
cat /sys/block/sda/queue/scheduler
The active scheduler will be enclosed in brackets. e.g. noop deadline [cfq]
For multiple drives use:
grep "" /sys/block/*/queue/scheduler
During the past month my backups failed (status interrupted) 1-2 times per week due to this memory leak.
When increasing heap size (node old space size) it takes longer but the backup fails when RAM usage eventually hits 100%.
I guess I’ll go with @Pilow ‘s workaround for now and create a cronjob for rebooting XO VM right before backups start.
Difficile de parler de « réalité » avec les benchmarks. Vérifiez aussi l'iodepth (qui dépend du type matériel que vous avez, sur du flash/NVMe vous pouvez monter à 128 ou 256), la latence entre en compte aussi bien sûr.
Le bottleneck principal est le monothreading de tapdisk, si vous testez sur plusieurs VMs différentes la somme va monter de manière assez régulière.
