samba - Mar 2007 - 3.0.24 weird errors, harmless?

A couple weeks ago I upgraded my servers from 3.0.14a to 3.0.24.  My
main server generally has about 800 to 1200 people connected to it at
any given time.  Its a Solaris 8/Sparc system, samba is configured as
security = domain .  Everything seems to be functioning fine, I haven't
noticed any odd behavior, haven't had any complaints.

But, a few weird errors have got me concerned.

1) I often run smbstatus, about once every dozen or so times of running
smbstatus I'll get an error like this..
tdb(/usr/local/samba/var/locks/locking.tdb): rec_read bad magic
0xd9fee666 at offset=250092, I run smbstatus again, literally 1 second
later and its fine.

2) Every several seconds to every several minutes depending, I guess
depending on how heavy the server's being worked I'm getting log.smbd
entries: 

[2007/03/07 08:16:57, 0] passdb/pdb_smbpasswd.c:startsmbfilepwent(204)
  startsmbfilepwent_internal: unable to open file
/usr/local/samba/private/smbpasswd. Error was
: Too many open files
[2007/03/07 08:16:57, 0]
passdb/pdb_smbpasswd.c:smbpasswd_getsampwsid(1378)
  Unable to open passdb database.

3) Again in log.smbd, I'm frequently seeing.

  Failed to set socket option SO_KEEPALIVE (Error Invalid argument)
[2007/03/07 08:44:15, 0] lib/util_sock.c:set_socket_options(261)
  Failed to set socket option TCP_NODELAY (Error Invalid argument)
[2007/03/07 08:44:15, 0] lib/util_sock.c:get_peer_addr(1229)
  getpeername failed. Error was Transport endpoint is not connected
[2007/03/07 08:44:15, 0] lib/util_sock.c:write_data(562)
  write_data: write failure in writing to client 134.124.32.148. Error
Broken pipe


I went back and looked at my old 3.0.14a logs, I wasn't getting any of
this stuff logged before the upgrade, wasn't having the smbstatus issue
either.

I've got another Samba server, using the identical compilation of Samba,
configured almost identically, it generally has 300 - 400 users
connected to it and I am not seeing any of these issues with it.

Tom Schaefer
University of Missouri Saint Louis

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Schaefer Jr, Thomas R. wrote:
> 2) Every several seconds to every several minutes depending, I guess
> depending on how heavy the server's being worked I'm getting
log.smbd
> entries: 
> 
> [2007/03/07 08:16:57, 0] passdb/pdb_smbpasswd.c:startsmbfilepwent(204)
>   startsmbfilepwent_internal: unable to open file
> /usr/local/samba/private/smbpasswd. Error was
> : Too many open files
> [2007/03/07 08:16:57, 0]
> passdb/pdb_smbpasswd.c:smbpasswd_getsampwsid(1378)
>   Unable to open passdb database.
IIRC, on Solaris, one would modify /etc/system to change the max number
of open files. There are probably other ways to do this too. What I
would do is rule out the obvious: that Samba may from time to time
exceed the number of allowable open files on your machine. I don't
recall the tools to count such things on Solaris or how to change them,
but the behavior is consistent with that problem (and thankfully, so are
the error messages!). You may never have had that happen on the old one
because of differences in the way the newer version acts (opens a few
more files now) or because your userbase has grown slightly to the limit.

- --
 ---- _  _ _  _ ___  _  _  _
 |Y#| |  | |\/| |  \ |\ |  | |Ryan Novosielski - Systems Programmer III
 |$&| |__| |  | |__/ | \| _| |novosirj@umdnj.edu - 973/972.0922 (2-0922)
 \__/ Univ. of Med. and Dent.|IST/AST - NJMS Medical Science Bldg - C630
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (MingW32)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFF9ws/mb+gadEcsb4RApGfAKDKMnPVtWnczS5ocLMJ34dAzQzpcACfVmxa
Z744YVTngxfyOwOe59+NJ9g=XpyL
-----END PGP SIGNATURE-----