[Bugs] [Bug 1287517] Memory leak in glusterd

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1287517



--- Comment #23 from Roman Tereshonkov <tereshonkov.roman at gmail.com> ---
The problem is reproducible for glusterfs 3.7.6 using the simple two nodes
cluster.
The proposed fix http://review.gluster.org/#/c/12927/ has no influence.

The setup:
1. any two nodes with network communication (node1, node2);
2. glusterfs 3.7.6 (./configure --prefix=/usr --libdir="/usr/lib64"
--sysconfdir=/etc --localstatedir=/var && make && make install)
3. rdma is not used (is it critical or not ?). glusterd.vol: option
transport-type socket
4. simple volume configuration for nodes:
     mkdir -p /brick/brick1
     gluster peer probe node1
     gluster volume create MYVOL replica 2 node1:/brick/brick1
node2:/brick/brick1 force
     gluster volume start MYVOL
     mount -t glusterfs node1:MYVOL /mnt/test (indeed I do not know if mounting
is needed or not to see the bug, not tested yet)
5. for ((N=0;N<20000;N++)) ; do gluster volume status all ; done
6. "top" to observe glusterd RSS memory growing (and it grows in both nodes)


NB: If you are using only one node with this setup the problem is not
reproduced.
    It look like two nodes communicating by network is the important
precondition.

-- 
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=LLJWZdHFct&a=cc_unsubscribe