Hi -
Long time Linux user, just upgraded from FC5 to FC6 and now can longer mount drives from my Solaris 8 (Sun Cluster 2.1) server. I can sucessfully mount filesystems from RedHat Enterprise Linux 4ES (Update 4) servers.
1) SELinux and IPtables are both disabled. 2) I can "ping" and "telnet" to the Solaris server, and "showmount -e" lists all the exported filesystems 3) tcpdump shows network traffic between the client and server 4) mount -t nfs server:/path /mnt commands time out with "Server not responding" messages
Any ideas?
Miner, Jonathan W (CSC) (US SSA) wrote:
Hi -
Long time Linux user, just upgraded from FC5 to FC6 and now can longer mount drives from my Solaris 8 (Sun Cluster 2.1) server. I can sucessfully mount filesystems from RedHat Enterprise Linux 4ES (Update 4) servers.
- SELinux and IPtables are both disabled.
- I can "ping" and "telnet" to the Solaris server, and "showmount -e" lists all the exported filesystems
- tcpdump shows network traffic between the client and server
- mount -t nfs server:/path /mnt commands time out with "Server not responding" messages
Any ideas?
Just for the record, FC6 works fine when mounting an export from a Solaris 10 (sparc) box here. Just a stab in the dark, but maybe it's related to NFS version conflicts? Have you done a sanity check to make sure other machines can mount the solaris export? You could try editing /etc/default/nfs on the solaris box, setting the min and max NFS versions to 3 and on your linux box mounting with the option -o vers=3
That's assuming you don't explicitly need NFS V2 or V4.
-----Original Message----- From: fedora-list-bounces@redhat.com on behalf of Ian Chapman Sent: Mon 10/30/2006 12:56 PM To: For users of Fedora Core releases Cc: Subject: Re: FC6 client fails to NFS mount from Solaris 8 server
Miner, Jonathan W (CSC) (US SSA) wrote:
Hi -
Long time Linux user, just upgraded from FC5 to FC6 and now can longer mount drives from my Solaris 8 (Sun Cluster 2.1) server. I can sucessfully mount filesystems from RedHat Enterprise Linux 4ES (Update 4) servers.
- SELinux and IPtables are both disabled.
- I can "ping" and "telnet" to the Solaris server, and "showmount -e" lists all the exported filesystems
- tcpdump shows network traffic between the client and server
- mount -t nfs server:/path /mnt commands time out with "Server not responding" messages
Any ideas?
Just for the record, FC6 works fine when mounting an export from a Solaris 10 (sparc) box here. Just a stab in the dark, but maybe it's related to NFS version conflicts? Have you done a sanity check to make sure other machines can mount the solaris export? You could try editing /etc/default/nfs on the solaris box, setting the min and max NFS versions to 3 and on your linux box mounting with the option -o vers=3
That's assuming you don't explicitly need NFS V2 or V4. --------------------------------
Yes, the Solaris cluster is working properly. It's the backbone of my infrastructure so I'd have much bigger problems if it failed! :) The workstation could mount from the cluster when running FC5, but after installing FC6 fails (with and without vers=3 option); however, I ran another test this morning, and FC6 will mount from a non-clustered Solaris 8 machine. Here is the output from the showmount and mount commands.
# uname -a Linux ac523421 2.6.18-1.2798.fc6 #1 SMP Mon Oct 16 14:37:32 EDT 2006 i686 i686 i386 GNU/Linux # showmount -e superfly Export list for superfly: /superfly/vol01 (everyone) /superfly/vol02 (everyone) /superfly/vol03 (everyone) /superfly/vol04 (everyone) /superfly/vol05 (everyone) /superfly/vol06 (everyone) /superfly/vol07 (everyone) /superfly/vol08 (everyone) # mount superfly:/superfly/vol01 /mnt mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (giving up). # mount -o vers=3 superfly:/superfly/vol01 /mnt mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (giving up).
I can generate a packet capture if that would be of interest.
Thanks
Adding to my earlier reply... see end of message
-----Original Message----- From: fedora-list-bounces@redhat.com on behalf of Ian Chapman Sent: Mon 10/30/2006 12:56 PM To: For users of Fedora Core releases Cc: Subject: Re: FC6 client fails to NFS mount from Solaris 8 server
Miner, Jonathan W (CSC) (US SSA) wrote:
Hi -
Long time Linux user, just upgraded from FC5 to FC6 and now can longer mount drives from my Solaris 8 (Sun Cluster 2.1) server. I can sucessfully mount filesystems from RedHat Enterprise Linux 4ES (Update 4) servers.
- SELinux and IPtables are both disabled.
- I can "ping" and "telnet" to the Solaris server, and "showmount -e" lists all the exported filesystems
- tcpdump shows network traffic between the client and server
- mount -t nfs server:/path /mnt commands time out with "Server not responding" messages
Any ideas?
Just for the record, FC6 works fine when mounting an export from a Solaris 10 (sparc) box here. Just a stab in the dark, but maybe it's related to NFS version conflicts? Have you done a sanity check to make sure other machines can mount the solaris export? You could try editing /etc/default/nfs on the solaris box, setting the min and max NFS versions to 3 and on your linux box mounting with the option -o vers=3
That's assuming you don't explicitly need NFS V2 or V4. --------------------------------
Yes, the Solaris cluster is working properly. It's the backbone of my infrastructure so I'd have much bigger problems if it failed! :) The workstation could mount from the cluster when running FC5, but after installing FC6 fails (with and without vers=3 option); however, I ran another test this morning, and FC6 will mount from a non-clustered Solaris 8 machine. Here is the output from the showmount and mount commands.
# uname -a Linux ac523421 2.6.18-1.2798.fc6 #1 SMP Mon Oct 16 14:37:32 EDT 2006 i686 i686 i386 GNU/Linux # showmount -e superfly Export list for superfly: /superfly/vol01 (everyone) /superfly/vol02 (everyone) /superfly/vol03 (everyone) /superfly/vol04 (everyone) /superfly/vol05 (everyone) /superfly/vol06 (everyone) /superfly/vol07 (everyone) /superfly/vol08 (everyone) # mount superfly:/superfly/vol01 /mnt mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (giving up). # mount -o vers=3 superfly:/superfly/vol01 /mnt mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (retrying). mount: mount to NFS server 'superfly' failed: timed out (giving up).
I can generate a packet capture if that would be of interest.
Thanks
After further testing, I've discovered that I can mount the filesystem from one of the cluster members. (The cluster is made of two nodes; fly_a and fly_b; the NFS service is known as "superfly" and its own IP address. Only one node is active at a time... currently fly_a is the active node)
# showmount -e fly_a Export list for fly_a: /superfly/vol01 (everyone) /superfly/vol02 (everyone) /superfly/vol03 (everyone) /superfly/vol04 (everyone) /superfly/vol05 (everyone) /superfly/vol06 (everyone) /superfly/vol07 (everyone) /superfly/vol08 (everyone) # mount fly_a:/superfly/vol01 /mnt Works!
Miner, Jonathan W (CSC) (US SSA) wrote:
Adding to my earlier reply... see end of message
-----Original Message----- From: fedora-list-bounces@redhat.com on behalf of Ian Chapman Sent: Mon 10/30/2006 12:56 PM To: For users of Fedora Core releases Cc: Subject: Re: FC6 client fails to NFS mount from Solaris 8 server
I haven't been following your thread very closely, but I came across this thread and wondered if it was related to your problem.
http://marc.theaimsgroup.com/?l=linux-kernel&m=116226648930820&w=2
-----Original Message----- From: fedora-list-bounces@redhat.com on behalf of Jay Cliburn Sent: Tue 10/31/2006 07:36 PM To: For users of Fedora Core releases Cc: Subject: Re: FC6 client fails to NFS mount from Solaris 8 Cluster
Miner, Jonathan W (CSC) (US SSA) wrote:
Adding to my earlier reply... see end of message
-----Original Message----- From: fedora-list-bounces@redhat.com on behalf of Ian Chapman Sent: Mon 10/30/2006 12:56 PM To: For users of Fedora Core releases Cc: Subject: Re: FC6 client fails to NFS mount from Solaris 8 server
I haven't been following your thread very closely, but I came across this thread and wondered if it was related to your problem.
http://marc.theaimsgroup.com/?l=linux-kernel&m=116226648930820&w=2 ---------------------
I don't think that is related. The patch appears to be related to UFS filesystems, not NFS.
Miner, Jonathan W (CSC) (US SSA) wrote:
I don't think that is related. The patch appears to be related to UFS filesystems, not NFS.
You're right, of course. I fixated on the Solaris angle and failed to see UFS versus NFS. Sorry.
I am having a similar problem, but with an IRIX nfs cluster, so hopefully we can triangulate.
The linux client is, of course, Fedora Core 6, all patched up to date.
The NFS server is part of a CXFS/Failsafe cluster, and I can mount shares on the linux system if I use the base hostname/IP address, but not when I attempt to mount via the IP alias address. The reason for the IP alias is to allow disconnect/relocation of cluster shares among cluster nodes (ie, move the filesystem export from one server to another)--that's failover.
I'm currently thinking it's a security "feature" in FC6. I'm trolling the newsgroups, looking at tcpdump/snoop and rpcinfo output.
Please add your comments to this bug:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=212471
So far the problem occurs with SUN, HP-UX, DEC Alpha and now IRIX clusters. It would be nice to see this resolved before it gets rolled into the mainstream RHEL release.
-----Original Message----- From: fedora-list-bounces@redhat.com on behalf of xianrice Sent: Sat 11/11/2006 03:58 AM To: fedora-list@redhat.com Cc: Subject: RE: FC6 client fails to NFS mount from Solaris 8 Cluster
I am having a similar problem, but with an IRIX nfs cluster, so hopefully we can triangulate.
The linux client is, of course, Fedora Core 6, all patched up to date.
The NFS server is part of a CXFS/Failsafe cluster, and I can mount shares on the linux system if I use the base hostname/IP address, but not when I attempt to mount via the IP alias address. The reason for the IP alias is to allow disconnect/relocation of cluster shares among cluster nodes (ie, move the filesystem export from one server to another)--that's failover.
I'm currently thinking it's a security "feature" in FC6. I'm trolling the newsgroups, looking at tcpdump/snoop and rpcinfo output.
Thanks a lot for the pointer--I've been dead in the water 'til now. I've added comments to the bug.
-
Ian Chapman -
Jay Cliburn -
Miner, Jonathan W (US SSA) -
potat0