Storage getting slow down between 00:00 and 00:30.

G

Gilberto Ferreira

Renowned Member
Aug 21, 2014
543
79
93
Joinville, Santa Catarina, Brazil, Brazil
www.konnectati.com.br
Hi there folks.

We have 3 Proxmox VE 9 nodes with share a HBA Storage. We don't have the storage model, because the provider didn't disclosure it.
The storage has a 6TB LUN in which we have an OCFS2 mounted.
During the day everything is completely normal.
We can create/destroy VM and take snapshot as well.
No slow down at all. Nothing.
But when clock hits midnight, everything goes down hill.
Even a single df command in Proxmox SHELL tooks some 5 secs to show the mounted points.
Backup task ran at 22h everyday.
FSTrim timer will never took 30 minutos to be complete.
I have another client with similar scenario and nothing happens like this one.
This problem has stopped for some day in the past, but are back again about 1 or 2 months.

So, I am in a dead end!
Anybody has some advice?

Thanks a lot
 
Last edited:
Look for a daily cronjob in /etc/cron.d/ and in /etc/cron.daily/ and run crontab -e (for the current user, usually root) and run systemctl list-timers ...
 
Well...
Turns out that after a month or so, the server provider just replaced the troublesome storage to a new one, and everything is fine. I've had hooked up the new HBA storage and format it as OCFS2 and everything is ok, at least in the storage standpoint.
Now, after move the disk for the bad storage to the new good one. some vms get FS errors, like FS turn itself into Read-only.
The reason for that, I guess, is that the VM stayed for so long in that bad storage, suffering from that weird problem between 00:00 and 00:30, that I thing this could potentialy generate some dirty blocks or badblocks in a way that, even in a new storage which is sound, now is getting FS error.
Some vm after I did fsck works fine.
So I really don't know if I am make any sense about that.
Any insight will be mostly appreciated.
Thanks
 
You must log in or register to reply here.
Top Bottom
Back