Quote:
Originally Posted by saiwang
I am checking a Linux VM unresponsive issue and I was told that it might be caused by some APP made Linux Kernel used too much memory,(more than 800MB on a 32bit system), I want to monitor memory usage of the Kernel, how can I do that? is the Kernel memory usage equal to the sum up of output of /proc/slabinfo?
|
Sounds like fishing in the dark. Go back to the person that suggested this and ask:
- on what do they base this assertion
- what evidence do they suggest you collect
There are many reasons a system might misbehave - get some evidence rather than jumping on what is probably the least likely. To satisfy your curiousity, use the numbers provided by /proc/meminfo rather than attempting to sum the slabs yourself - there are several fields that could be considered "kernel memory". And no, they aren't "all of it", but close enough to see trends in growth if any exists.
Personally I would be looking at constraints and device emulation imposed by your VMWare host (mentioned in your other thread, but left out of this).