Higher memory usage after moving to VZ 4.7

[javii]

Hi, I moved from a Centos 5.7 VZ 4.6 server to a Centos 6 VZ 4.7 server with same hardware specs.

I migrated containers to the new node and now I'm notice some containers are using lot of memory, I see the red icon on PIM. If I raise memory limit, containers are using even more memory and I see red icon again.

Anybody knows what could caused that and how to solve it?

Best regards.

[KristianM]

Hi,

when you moved the CTs from VZ 4.6 to VZ 4.7 did you actually also migrate the set memory limits from SLM to vSwap?
Did you actually use SLM or UBC before? With VZ 4.7 SLM is depricated and vSwap is the new memory management scheme used.

Regards,
Kristian

[javii]

I migrated the containers using PIM so I think it didn't changed anything.

Should I change containers to vSwap? How can I do that? Is it safe? It is on production.

Thank you!

[MakcyD]

Hi javii,

Yes, it is advised to change containers to vSwap.
This operation is safe - you just need to supply --ram and --swap parameters to vzctl set CTID --save command with corresponding values and these limits will be applied on the fly.

[stende]

Is there a safe way to do this in pva or pmc? It looks like those programs isn't upgraded for the vswap- system
I tried changing values with pva and got a mixed conf file that got parameteres sat so it wouldn't restart with the settings. So I had to edit the containers conf file and remove a lot of the old values that had been added,

[MakcyD]

You are partially right, PMC does not support vSwap as it was obsolete in favour of PVA.
But in case of PVA you should have latest PVA Agent installed on PVC node and PVA Management Node should be updated to build 4.6-1861 as well.

See the Knowlegde Base article to update instructions for both PVA Agent and PVA Management Node.