nmbd fails when run from xinetd
Hi,
I just had a long e-mail discussion about a problem with nmbd that I can't
solve and which the other person couldn't reproduce, because he is running
a bo system. Since the problem only exists with the 'hamm' version of the
samba package and not with the 'bo' version, I thought I'd post this
question here and not on debian-user. As soon as I know which program
fails here (nmbd, xinetd or both), i'll file a bug report (with Severity:
important, at least).
The problem is this: xinetd tries to start nmbd, which exits with an error
in the log file. Then nmbd is restarted by xinetd, exits with the same
error, etc., etc., until xinetd thinks nmbd is restarting too fast (which
doesn't happen on an otherwise loaded system, BTW) and disables the
service.
This is a typical /var/log/nmb, left by such an nmbd:
01/27/1998 02:25:16 netbios nameserver version 1.9.18alpha14 started
Copyright Andrew Tridgell 1994-1997
bind failed on port 137 socket_addr=130.89.222.95 (Address already in use)
make_subnet: Failed to open nmb socket on interface 130.89.222.95 for port 137. Error was Address already in use
ERROR: Failed when creating subnet lists. Exiting.
This happens on my hamm system, which is closely following all the
updates (from day to day, if possible). Thus, I have:
libc6_2.0.6-3
netbase_3.03-1
samba_1.9.18alpha14-1
xinetd_2.2.1-1
The above does NOT happen with any of the following:
- nmbd run from the command line
- nmbd run from inetd (and every other service run from xinetd)
- nmbd version 1.9.16p11-3 (the latest in bo) run from xinetd_2.2.1-1
- nmbd version 1.9.16p11-3 run from xinetd_2.1.7-3 (the latest in bo)
It DOES happen with:
- nmbd version 1.9.18alpha14-1 run from xinetd_2.2.1-1
- nmbd version 1.9.18alpha14-1 run from xinetd_2.1.7-3
I recall from memory that at least one of the samba_1.9.17p2* [1] packages
also worked with the xinetd that was in hamm at that time, but I can't
find any one of them for testing. I have been following hamm closely for
quite some time, but I can't recall exactly when nmbd started failing.
Now, does anybody know which program fails here? Or am I the only one with
these experiences? Can anyone reproduce this?
Remco
P.S. Hmm, after I changed this message somewhat to reflect some
last-minute testing, I think the problem lies with nmbd. Can anybody find
out what it is?
[1] I know this because I remember it fixed a security hole. I can find
the security hole in the changelog for 1.9.17p2-1, but the NMU that fixed
it first (by Joost, I believe) isn't mentioned. Is this a bug? Should
NMU's be mentioned in the changelog for 'normal' uploads? BTW, why isn't
this (libc5-based) NMU, which fixed a security hole, in bo?
--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-request@lists.debian.org .
Trouble? e-mail to templin@bucknell.edu .
Reply to: