samba - Aug 2000 - Oplock problem in 2.0.7 locks up Samba completely

first of all: thank you all for you excellent work on Samba.

However, I dare to complain about a problem with oplocks that of-
ten plagues the server here. The server is used  for  just  about
everything, Samba is used for file and print services but also as
a gateway to a Pervasive SQL database (via a named pipe).

Usually after a few days of perfect function (just over two weeks
at  the  last occurrence of the problem), all Samba services lock
up completely. The log is filled with messages like  the  follow-
ing:

[2000/08/14 09:49:14, 0] smbd/oplock.c:request_oplock_break(1204)
  request_oplock_break: no response received to oplock break request to pid 5060
on port 3708 for
dev = 3a01, inode = 16484
  for dev = 3a01, inode = 16484, tv_sec = 3997a286, tv_usec = c5bf0
[2000/08/14 09:49:46, 0] smbd/oplock.c:request_oplock_break(1204)
  request_oplock_break: no response received to oplock break request to pid 5060
on port 3708 for
dev = 3a01, inode = 16484
  for dev = 3a01, inode = 16484, tv_sec = 3997a286, tv_usec = c5bf0

The  system  runs a Linux 2.2.14-SMP kernel  on  a  recent  Intel
board  with  one  Xeon  550MHz processor and 256MB Ram. The share
filesystems are ext2 file sys- tems  on  a  lvm  managed  logical
volume.

In this situation, Samba must be stopped and restarted. It is not
clear  what causes the problem (if it were, this could be used as
a simple but very effective denial of service attack against Sam-
ba servers).

Even  if  the problem actually comes from some silly application,
it should not cause Samba to lock out all users! I  could  under-
stand two blocked users that compete for a lock, but not all.  Is
there a fix for this problem?

How can I further help to analyse the problem?

Mit herzlichem Gruss

                                        Andreas Mueller

PS: I tried to subscribe to the list, but did never get a confir-
    mation  from  listproc@samba.org. And the search engine seems
    to have been discontinued.  So please reply to me  as well as
    to the list, if you like.
--
Dr. Andreas Mueller        Beratung und Entwicklung
Bubental 53, 8852 Altendorf          afm@othello.ch
Tel: 055/4621483 Fax: 0554621485 Natel: 079/3556714

Dr Andreas F Muller <afm@othello.ch> wrote: 
| Usually after a few days of perfect function (just over two weeks
| at  the  last occurrence of the problem), all Samba services lock
| up completely. The log is filled with messages like  the  follow-
| ing:
| [2000/08/14 09:49:14, 0] smbd/oplock.c:request_oplock_break(1204)
|  request_oplock_break: no response received to oplock break request
to pid | 5060 on port 3708 for  dev = 3a01, inode = 16484 for dev 3a01, 
| inode = 16484, tv_sec = 3997a286, tv_usec = c5bf0 

As a pallitive, I'd put in 
	[globals]
		keepalive  = 60
to make samba check for failed clients every minute or so.

  This should make the smbd for the client that isn't responding
to clean up and exit.
  Note that this is not a fix: it's an avoidance procedure.

  The problem is probably at one of the clients: look for log messages
indicating which processes exited, and grep for those in the logs
to see which clients disconnected.
  Then make a client-specific conf file (with include = something.%m
see http://www.oreilly.com/catalog/samba/chapter/book/ch09_01.html)
and turn the client log level up to 3, log that to a seperate
file and look to see if anything obviously wrong shows up.

--dave
-- 
David Collier-Brown,  | Always do right. This will gratify some people
185 Ellerslie Ave.,   | and astonish the rest.        -- Mark Twain
Willowdale, Ontario   | //www.oreilly.com/catalog/samba/author.html
Work: (905) 415-2849 Home: (416) 223-8968 Email: davecb@canada.sun.com