PVE 6.4 - random reboots

[Fred Saunier]

Hello all,

We are experiencing what seem to be random reboots on our 2-node cluster.

Node 1 has 144 GB of RAM + 128 GB swap, and is hosting a single Windows VM with 128 GB of RAM (no ballooning).
Node 2 has 256 GB of RAM + 256 GB swap, and is hosting a single Windows VM with 230 GB of RAM (no ballooning). This VM replicates to Node 1 once a week.

On Node 2's last reboot, syslog shows :

systemd[1]: Started Proxmox VE replication runner.

This reboot happened on a Wed. at 10:00, when replication takes place on Fridays at 19:00.

Any help or pointers to troubleshoot this would be welcome, as adding swap did not seem to solve the reboots.

Thanks
Fred

 

[dcsapak]

did you enable ha?

also, since you mentioned replication, i guess you use zfs on those hosts?
maybe the hosts run out of memory and swapping is not fast enough? zfs by default can use up to 50% of the hosts memory.

here you can find info on how to limit zfs memory usage: https://pve.proxmox.com/wiki/ZFS_on_Linux#sysadmin_zfs_limit_memory_usage

otherwise, the log is maybe helpful or (if that does not show anything), setting up a remote syslog might catch some errors a local log does not

at last, i'd check the hardware, crashes are often caused by faulty/undersized hardware (e.g. faulty ram, bad psu, etc.)