Unfortunately no one can be told what fun_plug is - you have to see it for yourself.
You are not logged in.
Is there any way to get some more info into the fun_plug logfiles? I am having a lot of trouble with nfs and some info from unfsd could be really helpfull. So far unfsd has been working very good and the logfile is still empty after weeks of usage, not unusual to report I guess. But now something is really not working and the logfile is still empty. I could use all the info I can get.
Offline
krid wrote:
Is there any way to get some more info into the fun_plug logfiles? I am having a lot of trouble with nfs and some info from unfsd could be really helpfull. So far unfsd has been working very good and the logfile is still empty after weeks of usage, not unusual to report I guess. But now something is really not working and the logfile is still empty. I could use all the info I can get.
Start syslog to get the system log (in /var/log/messages). Afaik, there are two major reasons for unfs to fail: a) portmap not running or b) errors in exports file. "rpcinfo -p <dns-ip>" from a remote machine is a good test to see if NFS is running and responding. Good luck.
Offline
Syslogd is running, but unfsd doesn't log a lot. I restarted unfsd BTW. There is always this message when you restart unfsd that the portmapper is not killed. In the unfsd.sh is a: "killall portmap", I changed it to rpc.portmap and the portmapper is also restarted.
But now the nfs mounts which were working for weeks don't work anymore!
I get this in the logfile:
Dec 20 17:35:05 disk syslog.info syslogd started: BusyBox v1.6.1 Dec 20 17:35:26 disk daemon.warn portmap[17241]: connect from 127.0.0.1 to unset(100005): request from unprivileged port Dec 20 17:35:26 disk daemon.warn portmap[17242]: connect from 127.0.0.1 to unset(100005): request from unprivileged port Dec 20 17:35:26 disk daemon.warn portmap[17243]: connect from 127.0.0.1 to unset(100003): request from unprivileged port Dec 20 17:35:27 disk daemon.err portmap[17250]: cannot bind udp: Address already in use Dec 20 17:58:18 disk syslog.info syslogd exiting Dec 20 17:58:25 disk syslog.info syslogd started: BusyBox v1.6.1 Dec 20 17:59:34 disk daemon.err inetd[13701]: tftp/udp: bind: Address already in use
I get this answer from rpcinfo:
100000 2 tcp 111 portmapper
100000 2 udp 111 portmapper
100003 3 udp 2049 nfs
100003 3 tcp 2049 nfs
100005 1 udp 2049 mountd
100005 3 udp 2049 mountd
100005 1 tcp 2049 mountd
100005 3 tcp 2049 mountdOffline
NFS was not working was due to a typo somewhere in the export file. But there is nothing logged to the logfile. Is it possible to increase the verbosity of unfsd and/or rpc.portmap?
Offline
krid wrote:
NFS was not working was due to a typo somewhere in the export file. But there is nothing logged to the logfile. Is it possible to increase the verbosity of unfsd and/or rpc.portmap?
None that I know of. And to prevent such problems, the unfs start script creates the default exports file. Why did you change it?
Offline
I changed the exports file because I need something different then the defaults. I want to make a diskless server. I have the server, I had it booting with tftp, I only needed the root mounted over nfs. There are a lot of howto's about this subject, but not a lot of troubleshooting howto's.
A short overview of what I want over nfs:
root dir for diskless server over nfs,
some home directory's for users on the diskless server
a mythtv storage directory
slimserver music directory
and then some directory's over afp, but I might also do this with nfs. This is for a mac. With tiger unfs wasn't working, but maybe it does with leopard.
The reason I want this diskless server is that it will be a front and backend for mythtv. It is fanless and without disks I can put it anywhere I want because it is silent.
At this moment the DNS-323 is running with unfsd and afpd. So, on the current, not diskless, server I have mounted the slimserver directory and some data directory's. This works fine. In chrooted debian I have afpd running for a homedirectory on the mac and a timemachine backup directory. This also works fine. So, time to move on and get the diskless server running. There is so much to change that I will make a typo here an there and spent a night troubleshooting. It would be really helpfull to get as much info as possible in the logfiles.
Offline
Did you know that to mount /mnt/HD_a2/somedir, it's sufficient to export /mnt/HD_a2? There is no need to mount exactly the exported diretories. You can mount subdirs. So the default exports file should work fine in most cases.
Offline
fonz wrote:
Did you know that to mount /mnt/HD_a2/somedir, it's sufficient to export /mnt/HD_a2? There is no need to mount exactly the exported diretories. You can mount subdirs. So the default exports file should work fine in most cases.
Ok, didn't know that. But if I export the root file system I want the export to be as small as possible. I don't want to export an entire disk with the no_root_squash option.
I have this feeling that things might not work because of unfs. So, nfsmount from the initrd.img might not work with the nfs-user-server. I am installing another nfs-user-server on a different system to check this.
Last edited by krid (2007-12-21 22:11:03)
Offline
I tried unsf on a different system. This also doesn't work for nfsmount in initramfs. But, there seems to be a bug in the kernel nfs, which is fixed in 2.6.21. I have 2.6.19 so I have to make a new kernel first. But I don't want to stop the server right now. I would like to test if this problem has anything to do with userspace or kernel nfs. After that I can investate further.
Offline