From Bugzilla Helper: User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.7.6) Gecko/20050323 Firefox/1.0.2 Fedora/1.0.2-1.3.1 Description of problem: After updating util-linux to version util-linux-2.12a-21 on the clients mounting NFS-shares from the server (rh7.2) fails with following output: Mar 31 12:34:10 scheitan automount[5696]: mount(nfs): nfs: mount failure gargantua:/mnt/md0/home on /automount/home Mar 31 12:34:10 scheitan automount[5696]: failed to mount /automount/home Mar 31 12:34:13 scheitan automount[5698]: >> mount to NFS server 'gargantua' failed: server is down. There are no other log messages nor messages in the server's logs. Shared which were mounted before the update were usable but mounting additional shares failed. Reboot of client/server didn't change anything. rpcinfo and showmount returned normal autput as expacted. (showed available nfs ports and all exported dirs) After rollback to version util-linux-2.12a-19 everything is fine. Version-Release number of selected component (if applicable): util-linux-2.12a-21 How reproducible: Always Steps to Reproduce: 1. update util-linux to version 2.12a-19 2. try to mount NFS by UDP 3. read error message ;-) Actual Results: NFS was not mountable Expected Results: NFS share should have been mounted Additional info:
UDP mounts do work, but you need to use -udp explicitly as the default seems to have changed to TCP.
Adding udp option to my autofs config did it. But changing default options in such minor version changes is very dangerous, I think. (Some sort of pseudo bug ;-))
I have to agree. It has burnt us pretty nastily, our FC3 installations stopping to work randomly. Well, nothing against changing the default, but it should fall back to udp if tcp fails. I guess this can break a lot of network deployments.
(FWIW, now I see this is a dupe to bug 150775.)
Frank, please don't CLOSE this bug anymore.... it's important regression problem.
The error message I got from mount was mount to NFS server 'bfs' failed: server is down. This had me really confused for a while. If you cannot fail back to the old default the error message should at least allude to a possible solution. Somthing like "TCP mount failed. One of the people here found that SUSE seems to deal with things more cleanly. the SUSE mounter complained about TCP mounting but went on to use UDP.
Created attachment 112551 [details] Patch that fixes nfs mounts to rollback correctly The rpm and srpm that contains this patch is available at http://people.redhat.com/steved/bz152956
This is really a duplicate of #149798 which still lingers. Only in February I did not check that it affects not only x86_64 and kind of forgot about the issue.
'man 5 nfs' still claims that UDP is the default, by the way.
I tried the version util-linux-2.12a-22 from http://people.redhat.com/steved/bz152956/ and it seems to work well.
*** Bug 147017 has been marked as a duplicate of this bug. ***
*** Bug 136592 has been marked as a duplicate of this bug. ***
(In reply to comment #10) > I tried the version util-linux-2.12a-22 from > http://people.redhat.com/steved/bz152956/ and it seems to work well. I also tried util-linux-2.12a-22 from the above and it resolves my problems. This should be pushed out to the updates ASAP. The latest updates contains an X.org update and reboots will be much more common (and hitting this problem). I had to trace this problem back via my yum.log entries. Thanks for the effort.
*** Bug 153762 has been marked as a duplicate of this bug. ***