samba and inetd problems

[Ron Alexander]

Actually, nmbd seemed to be ok, it was smbd that messed up.

As of late Fri., we now know that the newly ported Spider Streams tcp code
did not implement SO_TYPE which is needed to allow Samba to determine if
fd[0] is a socket or stdin. Given that failure, each occurrence starts as a
daemon leading to all the other problems. We then implemented it, but when I
tested it there was still a bug (errno=5025). At that point, I could go no
further since everyone had left.

Ron

-----Original Message-----
From: samba-technical at samba.org [mailto:samba-technical at samba.org]On
Behalf Of James Sutherland
Sent: May 20, 2000 11:33 AM
To: Multiple recipients of list SAMBA-TECHNICAL
Subject: RE: samba and inetd problems


On Sun, 21 May 2000, Ron Alexander wrote:

(snip)
>
> As I pointed out in other posts, the real problem is with inetd. Remember
we
> are making VOS POSIX.1 compliant and porting streams tcp as well as samba,
> apache and gcc all at the same time. As soon as we get the bugs out of
> inetd, I am sure there will be no problem running smbd and nmbd from
inetd.

It seems there are problems running nmbd from inetd anyway, from what
others say. Running smbd should work OK, aside from performance issues,
though.

> At least on the Stratus, there is no problem with many processes using the
> same port. If that were the case, you could only have 1 telnet session.

That's not the point: the problem is with many process ACCEPTING
CONNECTIONS on the same port. You can have many telnet sessions open, but
can you have many telnet SERVERS running at once? If you do, which one
gets the next incoming session?


James.