[Bugs] [Bug 1457936] possible memory leak in glusterfsd with multiplexing

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1457936



--- Comment #7 from Raghavendra G <rgowdapp at redhat.com> ---
After changing code to not spawn so many threads per brick, I ran following
test (mount process, glusterd and server process all on same node):

* turned brick-mux on
* created 1000 volumes, each connecting to a single brick (1x1)
[root at unused shberry]# gluster volume info | grep "Volume Name" | wc -l
1000

* started all the 1000 volumes
* mounted all 1000 volumes simultaneously

Iteration 1:
============

[root at unused shberry]# mount | grep glusterfs | wc -l
1000

Following is the memory consumption I observed (from top) after all clients
were mounted:
 9608 root      20   0 82.4g 2.4g 3140 S  0.0 31.6   0:21.78 glusterfsd         
26323 root      20   0 1080m 272m 2604 S  0.0  3.5   1:45.59 glusterd           

Number of threads in server process:
[root at unused rhs-glusterfs]# ls -l /proc/9608/task | wc -l
2007

Iteration 2:
============

Memory consumption after unmounting and remounting all 1000 volumes:
 9608 root      20   0 82.4g 2.0g 2876 S  0.0 25.5   0:32.16 glusterfsd         
26323 root      20   0 1080m 218m 2548 S  0.0  2.8   1:46.26 glusterd           

Iteration 3:
============

[root at unused rhs-glusterfs]# ls -l /proc/9608/task | wc -l
2010

Top output:
 9608 root      20   0 82.4g 1.8g 2904 S  0.3 23.7   0:43.35 glusterfsd         
26323 root      20   0 1080m 212m 2548 S  0.0  2.7   1:46.95 glusterd           

So, looks like there is no leak during mount/umount operations.

I am going to repeat the same exercise tomorrow reverting all my code changes.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=HTG4nUI34u&a=cc_unsubscribe