single thread die?

[cerr]

Hi There,

I have a multi threaded logging application that's been deployed
already. Now upon deployment we now see that the server sometimes
stops working which i've never seen in our development
environment...?
Uhm when the logserver stops, the PID is stil l present and seems to
be a working process but i actually have to "kill -KILL" it in order
to get it back up. I'm wondering if a single thread can crash at all
(segmentation fault) or if it's "just" hanging somewhere. I thoughtin
case of a seg fault in a thread, it would pull down the whole process,
wouldn't it?
This is on a Linux system by the way.

Thanks for suggestions and hints!
Ron

 

[Keith Thompson]

I have a multi threaded logging application that's been deployed
already. Now upon deployment we now see that the server sometimes
stops working which i've never seen in our development
environment...? [...]
This is on a Linux system by the way.

Try either comp.programming.threads, comp.unix.programmer, or one of
the Linux groups.

 

[jacob navia]

cerr a écrit :

Hi There,

I have a multi threaded logging application that's been deployed
already. Now upon deployment we now see that the server sometimes
stops working which i've never seen in our development
environment...?
Uhm when the logserver stops, the PID is stil l present and seems to
be a working process but i actually have to "kill -KILL" it in order
to get it back up. I'm wondering if a single thread can crash at all
(segmentation fault) or if it's "just" hanging somewhere. I thoughtin
case of a seg fault in a thread, it would pull down the whole process,
wouldn't it?
This is on a Linux system by the way.

Thanks for suggestions and hints!
Ron

(1)
You need to attach a debugger to the running thread to see what it is doing,
if it is in an infinite loop or similar
(2) Maybe your application is deadlocked. Review your code.
(3) Try to see under what environment conditions your application stops.

In one word:

You have to debug it. This can be extremely complex and can take anywhere
between several miniutes to several years.

Good luck!

 

[cerr]

cerr a écrit :






(1)
You need to attach a debugger to the running thread to see what it is doing,
if it is in an infinite loop or similar
(2) Maybe your application is deadlocked. Review your code.
(3) Try to see under what environment conditions your application stops.

In one word:

You have to debug it. This can be extremely complex and can take anywhere
between several miniutes to several years.

Good luck!

Hah, that's encouraging! Today unfortunately is Friday the 13th too
but I'll try my luck (keep trying). I've looked at the code several
times (for weeks) already and can't seem to find the problem. The
painful thing is, that it happens only "every now and then" and then
again, it seems to run just fine for a few days and then BANG, it just
stops... I can't attach a debugger onto running system either...

Ron

 

[jacob navia]

cerr a écrit :

I can't attach a debugger onto running system either...

Ron

Why you can't? I am sure gdb can do that. Read the documentation!

Using GNU's GDB Debugger
Debugging A Running Process

http://dirac.org/linux/gdb/06-Debugging_A_Running_Process.php