Hi Gonçalo,
Gonçalo Borges wrote:
> Dear All...
>
> The glite-wms-workload is using 1.6GB of memory in my WMS. This is the
> 3rd time I see this behaviour in a 2 weeks period.
>
> top - 16:11:28 up 6 days, 2:43, 1 user, load average: 0.09, 0.24, 0.56
> Tasks: 129 total, 2 running, 127 sleeping, 0 stopped, 0 zombie
> Cpu(s): 0.2% us, 0.0% sy, 0.0% ni, 99.8% id, 0.0% wa, 0.0% hi,
> 0.0% si
> Mem: 2070188k total, 1969332k used, 100856k free, 42100k buffers
> Swap: 2096472k total, 1048160k used, 1048312k free, 138520k cached
>
> PID USER PR NI %CPU TIME+ %MEM VIRT RES SHR S
> COMMAND 3593 glite 20 0 0
> 2177:42 79.4 1699m 1.6g 4716 S glite-wms-workl
>
> I guess this is not normal... Can someone else check there values? Do I
> need to do some extra tunning?
>
> Cheers
> Goncalo
Here's a way I did it on our WMS (no compilation necessary):
1) Download HOARD library here:
http://www.cs.umass.edu/~emery/hoard/hoard-3.5.1/linux/libhoard.so
and save it in /usr/local/lib/
2) Modify /opt/glite/etc/init.d/glite-wms-wm file so that code after
"## NOTE" line looks like this:
## NOTE
if [ -f "/usr/local/lib/libhoard.so" ]
then
preload="LD_PRELOAD='/usr/local/lib/libhoard.so'"
fi
eval ${preload} ${binpath} --conf ${conf_file} --daemon ${pid_file} >
/dev/null 2>&1
3) Run /opt/glite/etc/init.d/glite-wms-wm restart
4) Check that the trick worked:
grep hoard /proc/`cat /var/glite/glite-wms-workload_manager.pid`/maps
should give you something like
009ef000-00a19000 r-xp 00000000 09:01 4395000 /usr/local/lib/libhoard.so
00a19000-00a1a000 rwxp 00029000 09:01 4395000 /usr/local/lib/libhoard.so
5) Enjoy :)
ps aux|grep workload_manager
glite 23902 20.9 7.7 438168 323540 ? Ssl 14:50 42:32
/opt/glite/bin/glite-wms-workload_manager --conf glite_wms.conf --daemon
/var/glite/glite-wms-workload_manager.pid
--
Cheers,
Andrey Kiryanov.
|