On 02/05/2012 07:14 PM, Emmanuel Dreyfus wrote:
Anand Avati<anand.avati@xxxxxxxxx> wrote:
It depends on the data set and translators loaded. On the server side
the default inode cache size is large enough to grow the process foot
print to a few hundered MBs. Having caching translators would increase
the usage further.
I get around 230 MB on the client. Is it reasonable, or is it a leak?
What kind of operations are performed on the client before it hits 230
MB? To figure out if it is a leak or not, you can use valgrind or make
use of the memory accounting infrastructure built in. To enable memory
accounting, you will need to build GlusterFS with -DDEBUG. Once memory
accounting is enabled, the accounting data can be viewed by triggering a
statedump.
Regards,
Vijay
