Well, here I run a web server and I do my backups over the network to the
Netware file server, because *IT* gets backed up to tape every night.
Well, I've noticed that when copying large amounts of data, after awhile
the 'cp' process goes to sleep (and it must be killed from another
terminal with kill -9) and then the NCPFS mount must be umounted and
remounted before it works again.
I have a backup script (a shell script) that will tar necessary info into
a .tar.gz file in /tmp, then copy that to the netware mount. Many times,
right near the end of the script, the cp process just goes to sleep.
Even if it doesnt, the next time around that I run the script (if
ihaven't umounted and remounted the netware drive) then the thing will
freeze up early on.
It seems that after copying large amounts of data, then the cp process
will go into sleep (in kernel 2.0.10 it would go into uninterupptable
sleep, so says 'ps',now it just is 'S')
I've included a sample transcript of the most recent incident, including
a tail of /var/log/syslog and /var/log/messages. I'm no kernel hacker,
so please let me know if there is anything I can do in order to provide
yall with more information -- like kernel profiling or anything (which I
know nothing about btw)
HERE IS THE TRANSCRIPT:
www:~$ ps aux |grep cp
root 615 0.3 0.7 812 248 p1 S 17:56 0:03 cp
/tmp/ehsu.tar.gz /
***(at this point blach did 'su' and killed the cp process)
/var/log/messages:
Jan 7 17:50:57 www backup: Beginning System Backup
Jan 7 17:50:57 www backup: Backing up /etc ...
Jan 7 17:50:59 www backup: Backing up Web Pages
Jan 7 17:51:08 www backup: Backing up CGI Scripts
Jan 7 17:51:08 www backup: Backing up Apache logs
Jan 7 17:51:13 www backup: Backing up Apache Configuration
Jan 7 17:51:28 www backup: Backing up /usr/archive
Jan 7 17:53:23 www backup: Backing up /home/blach
Jan 7 17:55:05 www backup: Backing up /home/ehsu ***(screws up here)
Jan 7 18:10:10 www su: blach on /dev/ttyp0
/var/log/syslog:
Jan 7 18:10:20 www kernel: ncp_rpc_call: send error = -512
Jan 7 18:10:20 www kernel: NCP: notice message: result=-11
Jan 7 18:10:20 www kernel: ncp_put_inode: could not close
www:~# ls -l /clinpro/web/page/backup/ehsu -l
/bin/ls: /clinpro/web/page/backup/ehsu: I/O error
www:~# /usr/local/bin/netware/ncpumount /clinpro
www:~# /etc/rc.d/rc.ipx
Configuring the IPX interface...
Mounting \\ACRCLAN\CLINPRO under /clinpro ... ok
www:~# tail /var/log/syslog |grep 18:
Jan 7 18:10:20 www kernel: ncp_rpc_call: send error = -512
Jan 7 18:10:20 www kernel: NCP: notice message: result=-11
Jan 7 18:10:20 www kernel: ncp_put_inode: could not close
Jan 7 18:20:21 www kernel: ncp_free_all_inodes: INODES LEFT!!!
Jan 7 18:20:21 www kernel: ncp_free_all_inodes: freeing inode
Jan 7 18:20:21 www last message repeated 19 times
www:~#
***There are several points to notice:
o The last three messages (18:20:21) come As as RESULT of umounting and
remounting
o Note I waited about 10 minutes after killing 'cp' before umounting and
remounting
Thank you VERY VERY VERY much for any help you can give me whether it be
a reply to linux-kernel, or a personal mail back to me. ( I don't
subscribe to the linware list; is it archived somewhere?)
I really need to get this fixed because .. well .. backups are an
important thing, especially for a Web site :)
Sincerely,
James S. Blachly
_
___(_)__ _ _ __ ___ ___
|___| / _` | ' \/ -_|_-<_
_/ \__,_|_|_|_\___/__(_)
|__/ -james.
James S. Blachly blach@www.acrc.uams.edu
Webmaster, Arkansas Cancer Research Center