High slab usage

[Macenger]

On a CentOS6.4 server with 6 GB memory, only 159 MB is free(including buffer and cache).

From top, it was identified that 1.2 Gb is used by a java process. To find out the remaining, the /proc/meminfo output was inspected. Surprisingly, almost 4 Gb was used by Slab, and that too "SUnreclaim"

Code: Select all

Slab:            4353792 kB
SReclaimable:      26104 kB
SUnreclaim:      4327688 kB

Upon inspecting the output of slabtop, it was identified that the slabs "size-256" and "size-192" were totally eating into the 4 Gb.

Is there a way to find out for which process those kernel objects are being used?

[gerald_clark]

6.4 is really old, unsupported and riddled with exploits.
Upgrade to the current 6.9 and retest.

[TrevorH]

There are more than 500 revisions of the CentOS 6 kernel from 6.4 to 6.9 and numerous bugs are fixed. Update.

[Macenger]

I agree CentOS6.4 have a lot of bug fixes.

However, I only need to steps or methods to identify process using a particular slab

[gerald_clark]

We cannot help troubleshoot unsupported systems.
Your problem may have fixed in one of the hundreds of updates since your install.
Update and retest.

[TrevorH]

That also doesn't mention the numerous security fixes that have been made since 6.4. At present your system is as secure as a swiss cheese... absolutely riddled with holes.