samba and inetd problems

[John E. Malmberg]

Ron Alexander <rcalex at home.company> wrote:

> Here is the snippet of code from the smbd pgm that shows that the pid file
> is NOT created if the process is not the daemon. Only the first invocation
> of smbd is the daemon and creates the pid file, all it's children are not
> daemon.
>
>   if (is_daemon) {
>     pidfile_create("smbd");
>   }
>
> I now know that when a copy of smbd is started via inetd, the check for a
> socket on fd[0] fails. As soon as the stcp guys find and fix that bug, I
> assume this problem will go away.

The difference is that on OpenVMS, all invocations of smbd are the daemon,
and there are no children.

The pidfile prevents all but the first invocation of smbd.

The following lines in config.h, and an additional module named
vms_pidfile_create.c that contains an empty pidfile_create() function.

#ifdef MOD_PIDFILE

 /* PID file concept doesn't work for OpenVMS */
#define pidfile_create samba_pidfile_create

#endif

For the nmbd, the shell script that launches it checks to make sure that it
is not already running.

-John
wb8tyw at qsl.network