Summary: | net-fs/nfs-utils-1.2.2-r2 prevents exports from being mounted by clients | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Andreas Proteus <proteuss> |
Component: | Current packages | Assignee: | Gentoo's Team for Core System packages <base-system> |
Status: | RESOLVED OBSOLETE | ||
Severity: | normal | CC: | flameeyes, proteuss |
Priority: | High | ||
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
Andreas Proteus
2010-08-02 08:38:46 UTC
I have discovered that net-fs/nfs-utils-1.2.2-r2 works if I restart the daemon with: /etc/init.d/nfs restart Please note however that upon rebooting the server the above error appears again only to go away if I manually restart the daemon. Okay, -r2 is mine so I should take the pieces... Is this with NFSv4 enabled? nfsv4 is enabled equery u nfs-utils * Found these USE flags for net-fs/nfs-utils-1.2.2-r2: U I - - caps : Use Linux capabilities library to control privilege + + elibc_glibc : ELIBC setting for systems that use the GNU C library - - ipv6 : Adds support for IP version 6 - - kerberos : Adds kerberos support - + nfsv3 : Enable support for NFSv3 - + nfsv4 : Enable support for NFSv4 + + tcpd : Adds support for TCP wrappers Okay... are you actually using it or was just the default? Because -r2 should only be fixing a couple of things for nfsv4+kerberos (which is not the case for you), while there is an actual possible trouble with the init script that was reported as bug #330795 but that shouldn't hit you since you don't have sec=sys... (In reply to comment #4) I tried: #export USE="-nfsv4" #emerge nfs-utils The problem is still there: Aug 2 14:52:42 olethros mountd[1376]: authenticated mount request from 192.168.1.6:764 for /mnt/library (/mnt/library) Aug 2 14:52:42 olethros mountd[1376]: qword_eol: fflush failed: errno 2 (No such file or directory) #/etc/init.d/nfs restart mounts ok, no error. And you say -r1 does not do that, after a reboot? Now that's strange because it has changed nothing between the two beside nfsv4. (In reply to comment #6) I had another look and it appears that -r1 suffers from the same problem. When I filed the bug report, after downgrading -r1 I restarted nfs manually (no reboot) and it worked. I erroneously assumed that -r1 it was doing the right thing and -r2 had the problem. My apologies. To sum up: When the system starts, although the exports appear to be available (showmount -e), the clients are unable to mount them, and the server reports an error as shown above. If nfs is restarted manualy (/etc/init.d/nfs restart) the problem goes away. The same happens with "/etc/init.d/nfs reload" or even with "exportfs -r". For the time being I circumvent this problem by adding the line: /usr/sbin/exportfs -r to /etc/conf.d/local.start ------------------------------------------------------ lastly, upon issuing: exportfs -r, it gives the curious error msg: exportfs: Warning: /mnt/library does not support NFS export But the exports are there and are mountable! Issuing exportfs -r for a second time there is no warning and everything is ok. Reassigning then. /mnt/library uses which filesystem? From mtab: /dev/sdb1 on /mnt/library type ext3 (rw,nosuid,nodev,noatime) Okay so "exportfs: Warning: /mnt/library does not support NFS export" is definitely wrong ^^;; (In reply to comment #10) Yes, its wrong but it appears to cause no harm and in spite of the warning the share can be mounted by the clients. It may be an indication as to what is causing this problem. This problem is not unknown. http://linux.derkeiler.com/Mailing-Lists/SuSE/2009-12/msg01140.html They do not offer any clues though. :-( nfs-utils-1.2.3 now in the tree if you want to retest |