Hi, I have a setup with 7 gluster nodes, 5 of them are running 3.2.5 and 2 are running 3.2.6. The storage is on two nodes of the 5 running 3.2.5. On the ones with 3.2.6 the memory usage of glusterfs is constantly growing, and reached 10GB in about two weeks while on the nodes with 3.2.5 it is steady between 300 and 500 MB. Let me know how to proceed further. Thanks, Calin
Hi Calin, Thanks for the info, and can you please let us know the general pattern of operations made on mount point? also the type of the volume (ie, gluster volume info).
Hi Amar, This is the mount point in question: localhost:/instances on /instances type nfs (rw,noatime,mountproto=tcp,intr,ac,acregmax=120,acdirmax=120,acregmin=10,acdirmin=60,nolock,addr=127.0.0.1,mountaddr=127.0.0.1) I use NFS over Gluster and mount it from localhost as described here: http://community.gluster.org/a/nfs-performance-with-fuse-client-redundancy/ It is used for serving web applications (some php files + static files). Here is the output of # gluster volume info Volume Name: instances Type: Replicate Status: Started Number of Bricks: 2 Transport-type: tcp Bricks: Brick1: data1-z1.presslabs.net:/bricks/instances Brick2: data2-z1.presslabs.net:/bricks/instances
Calin, Can you try our latest qa release http://bits.gluster.com/pub/gluster/glusterfs/src/glusterfs-3.3.0qa40.tar.gz if you are not yet on production? This would help us to see if this is already a fixed issue in upcoming 3.3.0 release.
(In reply to comment #3) > Calin, Can you try our latest qa release > http://bits.gluster.com/pub/gluster/glusterfs/src/glusterfs-3.3.0qa40.tar.gz if > you are not yet on production? This would help us to see if this is already a > fixed issue in upcoming 3.3.0 release. Unfortunately all the nodes are in production. Is anything I can do to track down these issues?
Calin Don, We made the release of GlusterFS-3.3.0, Please upgrade to that version which should fix your issue. Regards, Amar
Although I didn't upgrade to 3.3 yet, found that there were no issues with 3.2.6. The memory increase was caused by some problems with hostnames resolutions; the peers running 3.2.6 could not be resolved by the other peers in the network.