Hi All,

I've got a system which has been running various versions of slack for
over 3 years. Since upgrading 9.1 to 10.0, ntpd and samba have had
problems starting up via the rc.scripts.

They can be started manually from a bash session by root with no problems.

NTP: ntpd now starts up a second instance which hangs:
root 253 4.0 4.4 3516 3516 ? SLs 13:27 0:00 /usr/sbin/ntpd
root 256 0.0 4.4 3516 3516 ? S 13:27 0:00/usr/sbin/ntpd

This has been started in rc.local as usr/sbin/ntpd

Logfile says:

26 Sep 13:27:18 ntpd[253]: frequency initialized -56.738 PPM from
/etc/ntp/drift
26 Sep 13:27:18 ntpd[256]: signal_no_reset: signal 17 had flags 4000000
26 Sep 13:27:20 ntpd[256]: signal_no_reset: signal 14 had flags 4000000

SAMBA: nmdb now dies after 2 minutes. log.nmbd says:

[2004/09/26 13:27:16, 0] nmbd/nmbd.c:main(664)
Netbios nameserver version 3.0.5 started.
Copyright Andrew Tridgell and the Samba Team 1994-2004
[2004/09/26 13:29:31, 0] nmbd/nmbd.c:reload_interfaces(263)
reload_interfaces: No subnets to listen to. Shutting down...

smbd stays up. smb.conf checks OK.

Like I said before, these run fine when root starts them manually from a
bash shell. These problems didn't show up until the upgrade to 10.0

Thanks for any thoughts.

ScottJ

Scott Jordan <ta@sjordan.com> wrote:
> NTP: ntpd now starts up a second instance which hangs:
> root 253 4.0 4.4 3516 3516 ? SLs 13:27 0:00 /usr/sbin/ntpd
> root 256 0.0 4.4 3516 3516 ? S 13:27 0:00/usr/sbin/ntpd

I had a problem with a patched 8.1 system a few years ago -- I
had a cron job that did something relying on a flaky nfs mount;
somehow I ended up with about 8 (or more) instances of cron
processes.

Never found a solution, I assumed the nfs thing caused cron to
hang and somwhere along the line another instance got started
up. This is probably no help to you, but it is perhaps not
dissimilar.

#Paul

On 2004-09-26, Scott Jordan <ta@sjordan.com> wrote:


> This has been started in rc.local as usr/sbin/ntpd

Did you edit rc.local to start samba? That might be the problem. Slack 10
default is an empty rc.local file and samba being started by an rc.samba script
that starts during sysboot if it's executable. Root can also start, stop,
and restart using this script.

nb

In <pan.2004.09.26.19.34.16.28286@sjordan.com>, Scott Jordan wrote:

> Hi All,
>
> I've got a system which has been running various versions of slack for
> over 3 years. Since upgrading 9.1 to 10.0, ntpd and samba have had
> problems starting up via the rc.scripts.
> ...
> Thanks for any thoughts.
>
> ScottJ

Wild guess:
Maybe you have some left-over *.new files in your /etc/rc.d/ ??

Good luck
Franz


--
Franz M. Sauerzopf
Atominstitut, TU Wien

On Sun, 26 Sep 2004 19:34:17 +0000, Scott Jordan wrote:

> Hi All,
>
> I've got a system which has been running various versions of slack for
> over 3 years. Since upgrading 9.1 to 10.0, ntpd and samba have had
> problems starting up via the rc.scripts.
>
> They can be started manually from a bash session by root with no problems.
>
<snip>
> smbd stays up. smb.conf checks OK.
>
> Like I said before, these run fine when root starts them manually from a
> bash shell. These problems didn't show up until the upgrade to 10.0
>
> Thanks for any thoughts.
>
> ScottJ
>
Because you upgraded, I agree with the poster that says to check for
rc.*.new. Especially, rc.M, rc.inet1, rc.inet2, rc.hotplug.

The logged messages indicate networking hadn't been started when the
services were started. The services can be started manually later
because networking is started at some point, but there is a problem
with the sequence in the older scripts (a WAG).

--
Rot: 24 hours more, Joh Fredersen, and I will perfect the machine-man.
http://us.imdb.com/title/tt0017136/quotes

On Sun, 26 Sep 2004 19:34:17 +0000, Scott Jordan wrote:

> I've got a system which has been running various versions of slack for
> over 3 years. Since upgrading 9.1 to 10.0, ntpd and samba have had
> problems starting up via the rc.scripts.
<snip>

-Problems were self inflicted.

ntpd wasn't coming up because the internal dns cache came up after ntp
did. Apparently ntp 4.1 is more patient than 4.2. The 4.2 couldn't find
the server addresses as quickly as it wanted, then hung.

samba 3 did not like the 'bind interfaces only = yes' config instruction
which ran fine under samba 2.

Thanks to everyone for your inputs.

ScottJ