I've deployed glusterfs 3.2.0 in distributed replica mode, but it is very problematic. Every now and then, clients accessing certain files just hang. For example: md5sum /home/glusterfs/www/storage/article/2307/15064.jpeg The file is ~8k bytes big, so such operation should return almost immediately. It never returns. md5sum is just an example, but it causes huge problems to apache, which just hangs when trying to read such a file; number of processes grows, memory usage grows, until the server stops responding. The only way to "fix" this is to kill glusterfs and mount it again. Is it a known issue? I'm seeing it on all gluster clients, so it doesn't seem to be any isolated issue. It is a new 3.2.0 installation, not an upgrade. -- Tomasz Chmielewski http://wpkg.org
