bandwidth limit on manual bkp

It is not clear which bandwidth you mean. Bandwidth of the read-data-from-disk aspect, sending data from PVE to PBS, receiving data on PBS, or writing data on PBS to disk. Also it would have helped if you posted the actual command you are using...

One way to affect bandwidth for manual execution is to set a global limit for the used storage like "# pvesm set STORAGEID --bwlimit restore=KIBs" - taken from: "https://pve.proxmox.com/wiki/Backup_and_Restore --> "Bandwidth Limit"

For a "vzdump" you can use man vzdump:

Code: 
NAME
       vzdump - Backup Utility for VMs and Containers

       --bwlimit <integer> (0 - N) (default = 0)
           Limit I/O bandwidth (in KiB/s).
 
I want to limit the bandwidth when I start a bkp, because I have problems with IO delay when I do a bkp, so I assume that it is the speed of reading data from the disk on the node, because on the PBS I do not see anomalies and the IO delay is good, but on the node it goes up to 70%
 
Hi,
for backup, you can configure a node-wide limit in /etc/vzdump.conf with the bwlimit option. If you have issues with IO delay a good idea can also be to limit the amount of workers via the performance option, e.g.: performance: max-workers=2
 
The default value for max-workers is 16. Setting it to 2 is an example, finding the best value is something that depends on your setup and needs to be tested. The speed usually doesn't suffer even when choosing low values, but other users reported that it can be very helpful to reduce IO delay. I'd first try that setting, then with bandwidth limit. Again, what is actually helpful depends on the setup.
 
You must log in or register to reply here.
Top Bottom