Re: problem with getfh & nfs
Schlomo Schapiro wrote: > well, I am using the automounter and sometimes it gets stuck with stale > NFS handles. then I can't umount ("device busy"), can't remove the > automounter from the /proc/mount (though the process long died), and can't > even rmdir the mount point ("device busy"). > And then I really only can reboot to clean up this mess. I already tried > to go to single user and back to multi user, but it doesn't help. Have you tried mounting the file system (or instructing automounter to mount eh file system) with the "soft" option? I think this might solve this problem. Gilad. = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: problem with getfh & nfs
Hi, well, I am using the automounter and sometimes it gets stuck with stale NFS handles. then I can't umount ("device busy"), can't remove the automounter from the /proc/mount (though the process long died), and can't even rmdir the mount point ("device busy"). And then I really only can reboot to clean up this mess. I already tried to go to single user and back to multi user, but it doesn't help. Sincerely, Schlomo Schapiro --- email: [EMAIL PROTECTED] WWW: http://www.schapiro.org On Thu, 26 Oct 2000, Nadav Har'El wrote: > On Thu, Oct 26, 2000, Schlomo Schapiro wrote about "problem with getfh & nfs": > >.. > > Also, what can I do to get rid of stale NFS handles ? I mean besides > > rebooting the machine (which is not a Linux-style solution). > > I hope you'll get a better answer, but I just wanted to point out one thing: > one of the nice things about Linux (and Unix in general) is the decoupling > of the various parts of the system, and when one fails you can (at least in > priciple) restart it, without having to restart the whole system ("reboot"). > What Windows users know as "reboot the system for this tiny configuration > change to take effect" is known in Unix as "restart this one tiny service > for the changes to it to take effect". > > So if you're having problems with NFS onone filesystem, you can umount it, > and mount it again, and that should (minus any bugs) clear all the handles, > caches, and such stuff from the time of the old mount. > > [I know nfs is in the kernel now, but I hope that it still follows the general > Unix philosophy] > > > PS: Please reply to the list only > > > > PPS: List-master ! Please reinstate the Reply-To: header field ! > > ;) > As usual (without the reply-to) you'll get this reply to the list and to > yourself - pressing "g"(roup reply) is easier than having to cut and paste > the list address myself... > All you YAY-sayers out there: there's still a little time to vote! > > -- > Nadav Har'El |Thursday, Oct 26 2000, 27 Tishri 5761 > [EMAIL PROTECTED] |- > Phone: +972-53-245868, ICQ 13349191 |Lumber Cartel member #2224. > http://nadav.harel.org.il |http://come.to/the.lumber.cartel > = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: problem with getfh & nfs
Hi, Nadav! On Thu, Oct 26, 2000 at 10:16:26AM +0200, you wrote the following: > > PS: Please reply to the list only > > > > PPS: List-master ! Please reinstate the Reply-To: header field ! > > ;) > As usual (without the reply-to) you'll get this reply to the list and to > yourself - pressing "g"(roup reply) is easier than having to cut and paste > the list address myself... > All you YAY-sayers out there: there's still a little time to vote! Guys, if you're going to have a discussion about this at the end of each letter, please create a new mailing list for it. I'm getting sick of it already. We've created a poll, go state your opinion there. The majority will decide. So far it's 11 to 8 in favor of not having the header. The poll will be concluded tomorrow. Please quit bitching and moaning. P.S., FWIW: some e-mail programs, such as mutt, have a special feature to deal with this, and it works wonderfully for me. If you're having problems using this feature, ask for help on the mutt mailing list. In my opinion the MUA is the right place to solve this problem, and I expressed it via my vote on the poll. -- Alex Shnitman| http://www.debian.org [EMAIL PROTECTED], [EMAIL PROTECTED] +--- http://alexsh.hectic.netUIN 188956PGP key on web page E1 F2 7B 6C A0 31 80 28 63 B8 02 BA 65 C7 8B BA > *Real* programmers use > cat > a.out No way, /real/ programmers use zcat > a.out 'cause you can type faster that way. -- from a Slashdot discussion = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: problem with getfh & nfs
On Thu, 26 Oct 2000, Nadav Har'El wrote: > So if you're having problems with NFS on one filesystem, you can umount it, > and mount it again, and that should (minus any bugs) clear all the handles, > caches, and such stuff from the time of the old mount. ther is a problem with this solution: you need to check all dependant systems and stop them, so you could umount the file system (i.e. kill any process that holds files open on that file system) and after that you need to start the relevant processes again. another problem: if this file system is exported via NFS, and is currently mounted by a different machine, restarting the NFS server would (as far as i noticed) pause a few minutes of service disturbance to the remote machine (most likely freezing processes that try to access this file system, during that period). guy "For world domination - press 1, or dial 0, and please hold, for the creator." -- nob o. dy = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: problem with getfh & nfs
On Thu, Oct 26, 2000, Schlomo Schapiro wrote about "problem with getfh & nfs": >.. > Also, what can I do to get rid of stale NFS handles ? I mean besides > rebooting the machine (which is not a Linux-style solution). I hope you'll get a better answer, but I just wanted to point out one thing: one of the nice things about Linux (and Unix in general) is the decoupling of the various parts of the system, and when one fails you can (at least in priciple) restart it, without having to restart the whole system ("reboot"). What Windows users know as "reboot the system for this tiny configuration change to take effect" is known in Unix as "restart this one tiny service for the changes to it to take effect". So if you're having problems with NFS on one filesystem, you can umount it, and mount it again, and that should (minus any bugs) clear all the handles, caches, and such stuff from the time of the old mount. [I know nfs is in the kernel now, but I hope that it still follows the general Unix philosophy] > PS: Please reply to the list only > > PPS: List-master ! Please reinstate the Reply-To: header field ! ;) As usual (without the reply-to) you'll get this reply to the list and to yourself - pressing "g"(roup reply) is easier than having to cut and paste the list address myself... All you YAY-sayers out there: there's still a little time to vote! -- Nadav Har'El|Thursday, Oct 26 2000, 27 Tishri 5761 [EMAIL PROTECTED] |- Phone: +972-53-245868, ICQ 13349191 |Lumber Cartel member #2224. http://nadav.harel.org.il |http://come.to/the.lumber.cartel = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
problem with getfh & nfs
Hi friends, I get from time to time an error from my NFS server about getfh and nobody can mount anything. The server is quite heavily used and I was wondering wether it could be that the nfsd runs out of file handles ? Where in the kernel can I change the amount of file handles a process can get ? (Btw, samba also always complains about not enough file handles ...). Also, what can I do to get rid of stale NFS handles ? I mean besides rebooting the machine (which is not a Linux-style solution). Sincerely, Schlomo Schapiro PS: Please reply to the list only PPS: List-master ! Please reinstate the Reply-To: header field ! --- email: [EMAIL PROTECTED] WWW: http://www.schapiro.org = To unsubscribe, send mail to [EMAIL PROTECTED] with the word "unsubscribe" in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]