CEPH OSD consuming all memory

[Joakim Svensson]

Hello!

I have a a small CEPH cluster and one of the machines are using much more ram then the other. Every thing works fine except that all memory get consumed and then the all the swap. I have two identical machines, each with 12x2TB SATA disks and 40GB RAM and 8 core cpu. They have been in use for about ten months and up until now every thing has worked fine. They use to use around 80% of RAM but now one of them use 99% after an hour.

I have looked around and I can not understand why this is happening. Where would be the best place to start looking for a solution?

 

[Alwin]

Check the performance stats of your cluster nodes and go through the syslog/journal.

And please also have a look at our documentation.
https://pve.proxmox.com/pve-docs/chapter-pveceph.html#_precondition

 

[udo]

Hello!

I have a a small CEPH cluster and one of the machines are using much more ram then the other. Every thing works fine except that all memory get consumed and then the all the swap. I have two identical machines, each with 12x2TB SATA disks and 40GB RAM and 8 core cpu. They have been in use for about ten months and up until now every thing has worked fine. They use to use around 80% of RAM but now one of them use 99% after an hour.

I have looked around and I can not understand why this is happening. Where would be the best place to start looking for a solution?

Hi,
which versions you are running?
Which process need the most ram?

AFAIK there is/was an memory issue with ceph... get you the memory back if you restart the OSDs? (one by one)?

Udo