Red Hat Bugzilla – Bug 150775
NFSv2 UDP mounts broken in FC4test1
Last modified: 2007-11-30 17:11:01 EST
Trying to mount from a RHEL2.1 NFS server fails unless the specific mount options are given: this is a regression since FC3. [root@trash mnt]# mount file:/home scratch/ mount to NFS server 'file' failed: server is down. [root@trash mnt]# mount -onfsvers=2 file:/home scratch/ mount to NFS server 'file' failed: server is down. [root@trash mnt]# mount -onfsvers=2,udp file:/home scratch/ possibly-relevant message in /var/log/messages: Mar 10 14:11:55 trash automount[8382]: >> mount clntudp_create: RPC: Program not registered nfs-utils-1.0.7-1 kernel-2.6.11-1.1177_FC4
missed a line: the last mount command was the one which *did* work. [root@trash mnt]# mount -onfsvers=2,udp file:/home scratch/ [root@trash mnt]#
*** Bug 151107 has been marked as a duplicate of this bug. ***
This also effects FC3, after selinux-policy-targeted-1.17.30-2.93 The use of 'mount -oudp' is a successful work-around.
Sorry, there were a few more updates that happened today - util-linux.i386 2.12a-21 initscripts.i386 7.93.7-1
Oh yay, now my /home has gone away on my FC3 boxes too! This is a regression between: util-linux-2.12a-16 and util-linux-2.12a-21 reverting to the -16 util-linux fixes the problem. I didn't see a Test Update announcement go out for this, nor have I seen an Update announcement to fedora-announce-list.
It seems the specific regression introduced here by the patch to fix bug 141773 was already known for nearly two months. Why was that patch included in the FC3 update?
Adding back guilty party.
Changes between 2.12a-16 and 2.12a-21: * Fri Feb 25 2005 <SteveD@RedHat.com> 2.12a-20 - Changed nfsmount to only use reserve ports when necessary (bz# 141773) * Sat Dec 4 2004 Steve Dickson <SteveD@RedHat.com> 2.12a-19 - Corrected a buffer overflow problem with nfs mounts. (bz# 141733) * Mon Nov 29 2004 Steve Dickson <SteveD@RedHat.com> - Made NFS mounts adhere to the IP protocol if specified on command line as well as made NFS umounts adhere to the current IP protocol. Fix #140016 Next time, I will be supposed to test it in devel/ branch or release Test Update. re-assign back to Steve who probably better know where is a problem.
*** Bug 149790 has been marked as a duplicate of this bug. ***
Created attachment 112550 [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/bz150775
Steve's util-linux-2.12p-5 works for me for a Redhat-9 NFS server from both x86_64 and i686 clients. Thanks.
The 2.12p-5 package doesn't install for me on FC3, this is a destined-for-FC4 package? [root@pepsi ~]# rpm -Uvh util-linux-2.12p-5.i386.rpm error: Failed dependencies: /usr/bin/newgrp is needed by (installed) redhat-lsb-1.3-4.i386
newgrp has moved to the shadow-utils package, so there may need to be a simultaneous update of that package when the fix is released.
util-linux-2.12p is for FC4
Joe, try the one in http://people.redhat.com/steved/bz15295. that one should work with fc3.
http://people.redhat.com/steved/bz152956/ :-)
util-linux-2.12a-22 works for me against the RHEL2.1 NFS server.
util-linux-2.12a-22 also works for me on FC3 client connecting to Legacy RH 7.3 server (NFS v3)
We started getting "unable to mount, host is down" messages recently when trying to mount NFS shares, and after some googling came across the indicated bug. *However* installing the updated util-linux on our FC3 clients (to RedHat 9 server) hasn't fixed it, now I get : # rpm -q util-linux util-linux-2.12a-22 # mount -tnfs resserv:/installfiles /mnt/resserv mount to NFS server 'resserv' failed. Which is even more puzzling, because it's a mount _from_ resserv. Forcing nfsvers=3 (or 2) doesn't work either.
what is the output of "rpcinfo -p resserv" ?
# rpcinfo -p resserv program vers proto port 100000 2 tcp 111 portmapper 100000 2 udp 111 portmapper 100011 1 udp 864 rquotad 100011 2 udp 864 rquotad 100011 1 tcp 867 rquotad 100011 2 tcp 867 rquotad 100005 1 udp 56616 mountd 100005 1 tcp 51250 mountd 100005 2 udp 56616 mountd 100005 2 tcp 51250 mountd 100005 3 udp 56616 mountd 100005 3 tcp 51250 mountd 100024 1 udp 56617 status 100024 1 tcp 51252 status # rpm -qf `which rpcinfo` glibc-common-2.3.4-2.fc3.5 On the server box, all is well, btw: # mount -tnfs resserv:/installfiles /tmp/inst/ # ls /tmp/inst apache.org krb5-workstation-1.2.7-14.i386.rpm [...]
hmm... there should have been at least the following entries: 100003 2 udp 2049 nfs 100003 3 udp 2049 nfs 100003 2 tcp 2049 nfs 100003 3 tcp 2049 nfs It appears the nfsd daemons have died on your server. try doing a "service nfs restart" on the server to see if that helps. You also might want to see if /var/log/messages has anything in it that might explain why the nfsd died.
Oh, ignore me, I've rebooted the the NFS server now and all is well :-) Confirmed that this RPM fixes the issue with FC3 client, RH9 server. When is a proper fix going to be pushed out ?
Fixed and officially released in util-linux-2.12a-23. (today)