Problem with spin down

[jschmied]

I trying to spin down my hdd because they are idle 99,9% of the day.

If a backup job starts, the spin up takes to long and backup is failing consistently:

100: 2021-03-01 02:00:01 INFO: backup mode: snapshot
100: 2021-03-01 02:00:01 INFO: ionice priority: 7
100: 2021-03-01 02:00:01 INFO: creating Proxmox Backup Server archive 'vm/100/2021-03-01T01:00:01Z'
100: 2021-03-01 02:00:01 INFO: issuing guest-agent 'fs-freeze' command
100: 2021-03-01 02:00:23 INFO: issuing guest-agent 'fs-thaw' command
100: 2021-03-01 02:00:26 ERROR: VM 100 qmp command 'backup' failed - backup connect failed: command error: http upgrade request timed out
100: 2021-03-01 02:00:26 INFO: aborting backup job
100: 2021-03-01 02:00:26 ERROR: Backup of VM 100 failed - VM 100 qmp command 'backup' failed - backup connect failed: command error: http upgrade request timed out

Any ideas how to make the job waiting a bit longer? Otherwise I have to wake up the hdd's by a cron job a few minutes before backup starts.

 

[aaron]

Take a look at the hooks for backups ( https://pve.proxmox.com/pve-docs/chapter-vzdump.html#_hook_scripts). You could spin up the disks pre backup and wait a bit until they are up.

 

[jschmied]

If i understand it right, I have to use the hook at the PVE but this hook has to spin up disk at PBS.

The disks are starting at 0:00 due to some housekeeping jobs of PBS, as a simple solution I'll try to move backup jobs to 0:10

 

[aaron]

yes, the hook is run on the PVE node. But you could ssh into the PBS box.

 

[mar_565]

I trying to spin down my hdd because they are idle 99,9% of the day.

If a backup job starts, the spin up takes to long and backup is failing consistently:

Any ideas how to make the job waiting a bit longer? Otherwise I have to wake up the hdd's by a cron job a few minutes before backup starts.

How did you get it to work?
Im getting constant wirtes to the Datastorage Hdd (2048k), so it will not spin down.
PBS on the same machine as PVE, ext4 Datastorage.

 

[jschmied]

I have PBS on an extra machine. ZFS datastore is idle the whole day. It spins up at 0:00 for maintainance:


May 11 00:00:03 pbs systemd[1]: Starting Rotate log files...
May 11 00:00:03 pbs systemd[1]: Starting Daily man-db regeneration...
May 11 00:00:03 pbs rsyslogd[733]: [origin software="rsyslogd" swVersion="8.1901.0" x-pid="733" x-info="https://www.rsyslog.com"] rsyslogd was HUPed
May 11 00:00:03 pbs systemd[1]: logrotate.service: Succeeded.

so I scheduled the backups for 0:10.

The disks are sleeping from 0:30 til next midnight.
Root disk is a SSD.

I think having PVE and PBS on one machine defeats the reason for backup.
Datastore on PVE is accessed frequently from the VM's. Its SSD in my case.