From: Thomas Glanzmann on
Hello,
I have a inetd running on Solaris 10/sparc which crashes during startup:

Core was generated by `/usr/lib/inet/inetd start'.
Program terminated with signal 11, Segmentation fault.
#0 0x000196d4 in populate_defaults ()
(gdb) bt
#0 0x000196d4 in populate_defaults ()
#1 0x00019c88 in read_instance_cfg ()
#2 0x00015888 in handle_restarter_event ()
#3 0x00015d6c in process_restarter_event ()
#4 0x00017bb0 in event_loop ()
#5 0x00018150 in start_method ()
#6 0x00013ca0 in _start ()

any ideas?

Thomas
From: Thomas Glanzmann on
Hello,

* Thomas Glanzmann <thomas(a)glanzmann.de>:
> I have a inetd running on Solaris 10/sparc which crashes during startup:

> Core was generated by `/usr/lib/inet/inetd start'.
> Program terminated with signal 11, Segmentation fault.
> #0 0x000196d4 in populate_defaults ()
> (gdb) bt
> #0 0x000196d4 in populate_defaults ()
> #1 0x00019c88 in read_instance_cfg ()
> #2 0x00015888 in handle_restarter_event ()
> #3 0x00015d6c in process_restarter_event ()
> #4 0x00017bb0 in event_loop ()
> #5 0x00018150 in start_method ()
> #6 0x00013ca0 in _start ()

truss output is at: http://thomas.glanzmann.de/tmp/inetd.truss

Thomas
From: Gary Mills on
In <slrnhiv3ul.h6l.thomas(a)glanzmann.de> Thomas Glanzmann <thomas(a)glanzmann.de> writes:

>I have a inetd running on Solaris 10/sparc which crashes during startup:

> Core was generated by `/usr/lib/inet/inetd start'.
> Program terminated with signal 11, Segmentation fault.
> #0 0x000196d4 in populate_defaults ()
> (gdb) bt
> #0 0x000196d4 in populate_defaults ()
> #1 0x00019c88 in read_instance_cfg ()
> #2 0x00015888 in handle_restarter_event ()
> #3 0x00015d6c in process_restarter_event ()
> #4 0x00017bb0 in event_loop ()
> #5 0x00018150 in start_method ()
> #6 0x00013ca0 in _start ()

There should be further information in the messages log. The usual
problem is incorrect information in the
/var/svc/manifest/network/inetd.xml file. Similar core dump bugs
were recently fixed in opensolaris:

6713844 luupgrade skipped inetd.xml update in non-global zones
6613359 inetd dumps core if inetd_start method is not specified

--
-Gary Mills- -Unix Group- -Computer and Network Services-
From: Thomas Glanzmann on
Hello Garry,

> There should be further information in the messages log. The usual
> problem is incorrect information in the
> /var/svc/manifest/network/inetd.xml file. Similar core dump bugs were
> recently fixed in opensolaris:

[ Dec 21 16:43:04 Leaving maintenance because clear requested. ]
[ Dec 21 16:43:04 Enabled. ]
[ Dec 21 16:43:04 Executing start method ("/usr/lib/inet/inetd start") ]
[ Dec 21 16:43:04 Method "start" exited with status 0 ]
[ Dec 21 16:43:05 Stopping because process dumped core. ]
[ Dec 21 16:43:05 Executing stop method ("/usr/lib/inet/inetd stop") ]
[ Dec 21 16:43:05 Method "stop" exited with status 0 ]
....
[ Dec 21 16:43:13 Method "stop" exited with status 0 ]
[ Dec 21 16:43:13 Restarting too quickly, changing state to maintenance ]

> 6713844 luupgrade skipped inetd.xml update in non-global zones
> 6613359 inetd dumps core if inetd_start method is not specified

I looked at the bug reports and found nothing wrong.

I just copied the /var/svc/manifest/network/inetd.xml over from a different
machine and ran a vimdiff on it, the files are identical. I also ran:

svccfg import /var/svc/manifest/network/inetd.xml

I also tried to disable all the services list in

inetadm

And called

inetconv -i /etc/inet/inetd.conf

Nothing changed,

Thomas