Different Runtimes in VZDump runs:

[LA-Diego]

Hello, Good Day,

Version:
proxmox-ve: 5.3-1 (running kernel: 4.15.18-10-pve)
pve-manager: 5.3-9 (running version: 5.3-9/ba817b29)

Backup Disk mounted with this options:
(Target is in the same Layer2 Network, 1Gbit/s)
192.168.178.10:/PVE-VZDUMP-BACKUP on /mnt/pve/ergnas01 type nfs4 (rw,relatime,vers=4.2,rsize=131072,wsizeetrans=2,sec=sys,clientaddr=192.168.178.22,local_lock=none,addr=192.168.178.10)

Server was rebooted on 11.2.2019.
Backup is running every Sunday Evening.

I have the Problem that the Backup is running much faster when the Server is freshly restarted.
On 18.2 (one week after reboot) the Backup took 9:07:03 for 1403Gb. Another Week later the Backup took 11:10:01 for 1404Gb.

As you can see i the attached monitorng Graphs, there is a huge usuage of the Swap Memory during the Backup. I dont understand so much of System Design und how Memory/RAM works excatly, so i dont know if this is the source of the Problem.
Cant be normal the there is 2-3 hours difference between the Backup!?

Did you have a Solution or Suggestios for me ?

Thanks, Best Regards from Bavaria
Markus

 

[dietmar]

If there are many disk writes, backup slows down, and swapping may produce many writes ...

 

[LA-Diego]

Hello,

Thanks for your Reply. But why is it swappig so much during Backup ?

If the Disk Writes are the problem, why is the first backup after reboot much faster ?
I am looking for this Problem since Months, but cant find any useful tipp or idea...

 

[dietmar]

Thanks for your Reply. But why is it swappig so much during Backup ?

I have no idea what you run inside your VM ...

 

[LA-Diego]

Hello, Good Morning.

The Backup with VZDump is running on the PVE Host.
What is the relationship inside the VM and the VZDump on the Host?

Best Regards
Markus

 

[dietmar]

What is the relationship inside the VM and the VZDump on the Host?

The backup is basically a qemu/kbm block job ..

see https://git.proxmox.com/?p=pve-qemu...5250ca9f681e5377bd1821479259e81d4e78d;hb=HEAD