Red Hat Bugzilla – Full Text Bug Listing
|Summary:||openSUSE 11.3 failed to mount export from gluster-nfs server|
|Product:||[Community] GlusterFS||Reporter:||Bernard Li <bernard>|
|Component:||nfs||Assignee:||Shehjar Tikoo <shehjart>|
|Status:||CLOSED WONTFIX||QA Contact:|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
Description Bernard Li 2010-09-23 19:18:58 EDT
When I tried to mount a gluster volume exported via gluster-nfs using the following command, it would hang and eventually timeout: # mount gluster-nfs:/export /share However, if I specify the nfsvers option, it works: # mount -o nfsvers=3 gluster-nfs:/export /share Perhaps there is some issue with NFS version detection? Running nfs_beta_rc14 on the server, and nfs-client-1.2.1-7.2 on openSUSE 11.3 client.
Comment 1 Shehjar Tikoo 2010-09-23 23:46:32 EDT
What kernel version does distro have? Recent kernels have moved the default nfs mount versions to 4 instead of 3.
Comment 2 Bernard Li 2010-09-24 00:40:21 EDT
Kernel version is 2.6.34-12-desktop and you're right, it also times out if I execute the following: # mount -o nfsvers=4 gluster-nfs:/export /share I found this bugzilla entry for openSUSE 11.3 regarding NFS version fallback: https://bugzilla.novell.com/show_bug.cgi?id=557138 Will do more troubleshooting and report back tomorrow. Thanks.
Comment 3 Bernard Li 2010-09-24 14:03:22 EDT
I have updated my nfs-client to 1.2.1-34.2 (from openSUSE Factory) which supposedly has the above openSUSE bug 557138 fixed. However, it still won't fall back from NFSv4 to NFSv3. What does gluster-nfs return when a mount request fails? ENOENT or EPERM or neither? Please let me know if I can provide any additional information to fix this issue, thanks!
Comment 4 Shehjar Tikoo 2010-09-25 02:41:50 EDT
(In reply to comment #3) > I have updated my nfs-client to 1.2.1-34.2 (from openSUSE Factory) which > supposedly has the above openSUSE bug 557138 fixed. However, it still won't > fall back from NFSv4 to NFSv3. > > What does gluster-nfs return when a mount request fails? ENOENT or EPERM or > neither? I see the negotiation depends on receiving a ENOENT from the server but because we dont support nfsv4, we dont register a NFSv4 service with portmapper. After that, portmap does not return a NFSv4 port to nfsv4 clients trying to connect. So a timeout occurs. > > Please let me know if I can provide any additional information to fix this > issue, thanks!
Comment 5 Shehjar Tikoo 2010-10-05 04:43:48 EDT
Closing. There is nothing that can be done here because we dont register a NFSv4 program with the portmapper, so the question about returning a ENOENT or EPERM for a NFSv3 export doesnt arise. I may consider registering a fake nfsv4 program just to return a ENOENT to be nfs client friendly but that will prevent a genuine nfs v4 server from registering on the same machine.