Description of problem: As many know, and more importantly many do not know, localhost (loopback) NFS mounts can lead to deadlocks under heavy load. There should be a warning in the documentation regarding this bug. Although it is NOT a GFS bug it can and will cause problems for those without the knowledge that I learned the hard way today on a production server. Version-Release number of selected component (if applicable): I would imagine this applies to all versions until the kernel is patched. How reproducible: mount a localhost with nfs and rsync ca. 10+ GB files (not sure what the lower limit is and I tried ca 11 GB when it deadlocked) Steps to Reproduce: 1. mount local volume per nfs 2. copy files 3. wait for deadlock and recover the server while hoping for no major damage Actual results: deadlock Expected results: no deadlock Additional info: I was thinking the warning could be added here: https://github.com/gluster/glusterfs/blob/master/doc/admin-guide/en-US/markdown/admin_settingup_clients.md#nfs
http://lwn.net/Articles/595652/ contains a fairly complete description. Note that the changes mentioned in the article can be made for the Linux kernel NFS-server, but not for an NFS-server in user-space.
This bug is getting closed because the 3.5 is marked End-Of-Life. There will be no further updates to this version. Please open a new bug against a version that still receives bugfixes if you are still facing this issue in a more current release.