Problem: RDB write waiting time increased to 1000ms since upgrade to PVE7

[KaiS]

I updated from PVE 5 to PVE6 to PVE 7 last weekend, and from CEPH Luminous to Nautilus to Octopus (in the right order).

Now i mentioned some warnings in my zabbix server monitoring for one node:



Seems, that write response/waiting time on some rdb devices are going up or somthing is ""hanging"

Look at this:



Yesterday at 19:00 I did the last step upgrading from PVE6 to PVE7.
I don´t upgrade to Ceph Pacific

The write waiting time for some RDB devices is going up from 50ms to 1000ms.

Oher devices on same node don´t have any problem:


I have no clue, how to fix this. And no clue how to find the rdb devices, because zabbix named it as "rdb15" and not with the device name used by proxmox.

So please, can someone help me to find/identify the affected devices. And maby a hint how to solve this problem?

All other devices seems to work fine.

Thanks,
Kai

 

[mira]

Please provide the output of ceph -s and ceph health detail.
What does the `Apply` and `Commit` Latency in the GUI show? Node -> Ceph -> OSD


What's the output of rbd showmapped?

 

[KaiS]

Hi Mira,

thanks for your reply. Your "rbd showmapped" command helped me a lot. So I had identified the related disks. I backuped the VM´s with this disks, deleted them and reimport VM from backup. Now it seems my problem is gone.

Best regards,
Kai