Hi!
I’ve just found my server completely offline (bad moment with the phone almost discharged and sync stopped). Accessed via IPMI and it was frozen. Just a power reset worked.
It looks like everything is fine now, but how to start diagnosing what happened?
These messages were in the login screen but I’m not sure when they appeared:
It’s running Proxmox 5.3-5
is your server using ZFS ?
Yes
might be worthwhile to first investigate the “txg_sync” blocked error in relation to ZFS (just a guess, I don’t have any specific experience, but that’s where I might start)
also, if you have a /var/log/kern.log file (also check for previously saved rolled-over logs) then that might be useful to get more easy-to-read timestamps for the same error messages
I would look at limiting the size of your ARC. If you don’t it kinda spins outta control and eats itself alive.
@harambe may be able to offer some insight.
Francisco
Haven’t seen that one before… double check the drive health first. One of those might be starting to go if it’s hanging.
Beyond that, capping the ARC can’t hurt if you’re running it at defaults (it’ll eat up to 50% of ram by default). ZFS on Linux - Proxmox VE - general rule is like 1GB ram per TB of usable storage. I have my boxes with small SSDs/HDDs at 4-8GB max.
I’d also upgrade to PVE6 & latest zfs on linux before doing anything drastic. Upgrades have gone smooth for me.
