-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 20/09/12 21:37, Guillaume Rousse wrote: > Le 20/09/2012 20:17, Anne Wilson a écrit : >>>> Looks as though the problem relates to rpc on the server. I >>>> don't mind doing the digging, but can you point me to the >>>> right place to start the hole? >>> I'd try to understand why the nfs daemon (rpc.nfsd) is >>> listening on IPv6 only, whereas the mount daemon (rpc.mountd) >>> is listening both on IPv4 and IPv6. >>> >> Where is that configured? > There is no such 'use IPv6 only' configuration option AFAIK, this > seems rather like a problem. Maybe not specific to NFS, BTW. > > There is no --verbose flag for rpc.nfsd, but you may use rpcdebug > instead: rpcdebug -m nfsd -s all > > Then, restart the nfs-server service, and check your logs. > Sep 21 19:18:05 borg2 nfs-server[4320]: /etc/sysconfig/nfs-server: line 11: -d: command not found Sep 21 19:18:05 borg2 nfs-server[4320]: Stopping NFS kernel daemon Sep 21 19:18:05 borg2 rpc.mountd[1765]: Caught signal 15, un-registering and exiting. Sep 21 19:18:05 borg2 nfs-server[4320]: Stopping rpc.mountd:[ OK ] Sep 21 19:18:08 borg2 kernel: [110004.094558] nfsd: freeing readahead buffers. Sep 21 19:18:08 borg2 kernel: [110004.095309] nfsd: last server has exited, flushing export cache Sep 21 19:18:08 borg2 nfs-server[4320]: Stopping nfsd:[ OK ] Sep 21 19:18:08 borg2 nfs-server[4320]: Unexporting directories for NFS kernel daemon... Sep 21 19:18:08 borg2 nfs-server[4359]: /etc/sysconfig/nfs-server: line 11: -d: command not found Sep 21 19:18:08 borg2 nfs-server[4359]: Exporting directories for NFS kernel daemon... Sep 21 19:18:08 borg2 nfs-server[4359]: Starting NFS kernel daemon Sep 21 19:18:08 borg2 nfs-server[4359]: Starting nfsdrpc.nfsd: unable to bind inet TCP socket: errno 98 (Address already in use) Sep 21 19:18:08 borg2 kernel: [110004.460625] set_max_drc nfsd_drc_max_mem 765952 Sep 21 19:18:08 borg2 kernel: [110004.461595] nfsd: creating service Sep 21 19:18:08 borg2 kernel: [110004.461601] nfsd: allocating 32 readahead buffers. Sep 21 19:18:08 borg2 kernel: [110004.461624] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory Sep 21 19:18:08 borg2 kernel: [110004.461677] NFSD: starting 90-second grace period Sep 21 19:18:08 borg2 nfs-server[4359]: [ OK ] Sep 21 19:18:09 borg2 nfs-server[4359]: Reloading rpc.idmapd[ OK ] Sep 21 19:18:09 borg2 rpc.mountd[4403]: Version 1.2.5 starting Sep 21 19:18:09 borg2 nfs-server[4359]: Starting rpc.mountd[ OK ] Sep 21 19:19:39 borg2 kernel: [110094.560075] NFSD: laundromat service - - starting Sep 21 19:19:39 borg2 kernel: [110094.560086] NFSD: end of grace period Sep 21 19:19:39 borg2 kernel: [110094.580278] NFSD: laundromat_main - sleeping for 90 seconds Sep 21 19:21:06 borg2 dovecot: imap-login: Disconnected (no auth attempts): rip=192.168.0.40, lip=192.168.0.40, TLS handshaking: Disconnected Sep 21 19:21:09 borg2 kernel: [110184.800046] NFSD: laundromat service - - starting Sep 21 19:21:09 borg2 kernel: [110184.800060] NFSD: laundromat_main - sleeping for 90 seconds Sep 21 19:22:39 borg2 kernel: [110275.040069] NFSD: laundromat service - - starting Sep 21 19:22:39 borg2 kernel: [110275.040083] NFSD: laundromat_main - sleeping for 90 seconds Sep 21 19:24:09 borg2 kernel: [110365.280038] NFSD: laundromat service - - starting Sep 21 19:24:09 borg2 kernel: [110365.280047] NFSD: laundromat_main - sleeping for 90 seconds
The Line 11 that it complains about says RPCMOUNTD_OPTIONS="--port 4003" -d I *don't think* I edited that, but after all the reading around I've done it is just possible that I did. I do feel, though that I saw that port mentioned and felt that I shouldn't edit it, since I haven't a clue which of the many ports mentioned in the logs is the one it's expecting. Sorry I can be more specific. I've removed the '-d' and restarted. Sep 21 19:30:06 borg2 nfs-server[5169]: Stopping NFS kernel daemon Sep 21 19:30:06 borg2 rpc.mountd[4403]: Caught signal 15, un-registering and exiting. Sep 21 19:30:06 borg2 nfs-server[5169]: Stopping rpc.mountd:[ OK ] Sep 21 19:30:09 borg2 kernel: [110725.276891] nfsd: freeing readahead buffers. Sep 21 19:30:09 borg2 kernel: [110725.277654] nfsd: last server has exited, flushing export cache Sep 21 19:30:09 borg2 nfs-server[5169]: Stopping nfsd:[ OK ] Sep 21 19:30:09 borg2 nfs-server[5169]: Unexporting directories for NFS kernel daemon... Sep 21 19:30:10 borg2 nfs-server[5208]: Exporting directories for NFS kernel daemon... Sep 21 19:30:10 borg2 nfs-server[5208]: Starting NFS kernel daemon Sep 21 19:30:10 borg2 nfs-server[5208]: Starting nfsdrpc.nfsd: unable to bind inet TCP socket: errno 98 (Address already in use) Sep 21 19:30:10 borg2 kernel: [110725.681229] set_max_drc nfsd_drc_max_mem 765952 Sep 21 19:30:10 borg2 kernel: [110725.683544] nfsd: creating service Sep 21 19:30:10 borg2 kernel: [110725.683551] nfsd: allocating 32 readahead buffers. Sep 21 19:30:10 borg2 kernel: [110725.683569] NFSD: Using /var/lib/nfs/v4recovery as the NFSv4 state recovery directory Sep 21 19:30:10 borg2 kernel: [110725.683610] NFSD: starting 90-second grace period Sep 21 19:30:10 borg2 nfs-server[5208]: [ OK ] Sep 21 19:30:10 borg2 nfs-server[5208]: Reloading rpc.idmapd[ OK ] Sep 21 19:30:10 borg2 rpc.mountd[5251]: Version 1.2.5 starting Sep 21 19:30:10 borg2 nfs-server[5208]: Starting rpc.mountd[ OK ] That looks better, I think? However, 'mount -a' on the client still doesn't connect. >>> BTW, you're also lacking rpc.idmapd, which is mandatory for >>> NFSv4 support, but that's a secondary issue. >>> >> I've done a lot of googling, and most of what I found was >> ubuntu-specific. However, I found a recent entry saying that it >> is provided by nfs-utils - that that is installed. nfs-common >> and nfs-server are both running, according to mcc. > nfs-common service, as 90% of similar services, use an > /etc/sysconfig/nfs-common congiguration file. Just set NEEDIDMAPD > to yes to ensure it runs. > > Also, you'd better forget GUIs when trying to debug issue, and > ressort to command line instead, you'll have much more control and > feedback. > That's fine. I prefer the command line where it is an area that I've used before, but I'm nervous of it when I don't really know what I'm doing, and assume that MCC will safeguard against my errors. >> Following another lead, I tried to locate nfsmount.conf and >> found that I had a backup containing it, and a man page, but no >> actual file. > That's an interesting point, it means your actual setup differs > from the default package one. Try 'rpm --verify nfs-utils' and > 'rpm --verify nfs-utils-client' to check what has been modified > exactly. > >> This is incredibly difficult. Users shouldn't have to jump >> through this number of hoops to do something so basic to >> productivity :-( > I'm not sure what users should have to do or not, but I'm quite > sure this kind of imprecations won't help you solve the issue. > Of course :-) but you have to admit it was a pretty mild statement of my frustration :-) > If everything else fails, you may also inhibit IPv6 completly: I doubt if that would cause any problems - at least not for some time yet. > just add 'install ipv6 /bin/true' in any file under > /etc/modprobe.d, and reboot. > Really? That sounds counter-intuitive. Can you explain? And if I inhibit IPv6 completely, will that automatically enable IPv4, or am I still looking for the reason for that lack? @AL13N - > maybe those services were started when there was no ipv4 set up > yet, and thus was ignored? I hesitate to say that anything is impossible, but this was a clean, bog-standard Mageia2 install. I can't think why that should be so, can you? This is seriously impeding my work, but at least I'll have learned from it :-) Anne -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.12 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/ iEYEARECAAYFAlBctTMACgkQj93fyh4cnBf0sQCdGDjFZcuk7dlgJx4drpACNP3c sfcAn3oe1iI5Ny0rNX1qGSX0rvS3qm7R =0Bbq -----END PGP SIGNATURE-----
