memory leak?

[Ivan Voras]

On 08/23/10 11:50, n dhert wrote:
> After a reboot of my FreeBSD 8.0-p4 system
> a vmstat shows:
> Mon Aug 23 08:40:00 CEST 2010
>  procs      memory      page                    disks     faults         cpu
>  r b w     avm    fre   flt  re  pi  po    fr  sr da0 pa0   in   sy   cs us
> sy id
>  1 1 0   1515M  7118M   816   5   5   0   726   0   0   0  138 2376 1740  1
> 2 97
> My system had been allocated 8 Gb memory in a Vmware ESXi4 virtual machine.
> avm = active virtual pages,  fre = size of free list
> 2525M + 7118M is 9600M  is this normal? (more than 8 Gb..)

See here, on a physical machine with 4 GB RAM:

 procs      memory      page                    disks     faults         cpu
 r b w     avm    fre   flt  re  pi  po    fr  sr ad0 cd0   in   sy   cs
us sy id
 1 0 0   5571M  1254M    30   0   0   0    29   0   0   0    8  519  522
 0  0 100

We have 5571+1254=6825 MB.

The catch is in the description - active *virtual* pages belonging to
processes - they do not have to correspond to physical RAM.

> six days ago is was:

>  3 1 0   2446M  1552M   628   0   0   0   621   6   0   0   30  939  572  1

>  5 22 0   2260M   816M   489   0   0   0   478   2   0   0   27  957  580

> avm + free adds up to only 3 Gb.

> Where is the rest of the memory ??

There are a lot other uses for memory. Try running "top" a couple of
time when you seem to detect the leak and add together the various
categories of memory.

> Is there a memory leak, how to find what causes the memory leak?

If indeed there is, which is not probable, you first need to discover is
it in a userland program or in kernel. If in kernel, use "vmstat -m" to
track usage over time and see if some category increases when it shouldn't.

Also, since you are running vmware, maybe you should try running
emulators/open-vm-tools and its vmmemctl driver - it could help if your
host is running out of memory.