max threads blown?

Hi,

A process has thrown sig 11 on a pthread_create but I don't belive at
the moment that the process itself blew it's thread limit (256).

I am guessing there may be an issue with max threads at the kernel
level. I need to find out how many threads are active on the system in
total. Is there a utility to do this (easily)?


Kernel values are:
nkthread 2048 2048
max_thread_proc 256 256


Adam.

Adam Skeggs wrote:

> Hi,
>
> A process has thrown sig 11 on a pthread_create but I don't belive at
> the moment that the process itself blew it's thread limit (256).
>
> I am guessing there may be an issue with max threads at the kernel
> level. I need to find out how many threads are active on the system in
> total. Is there a utility to do this (easily)?
>
>
> Kernel values are:
> nkthread 2048 2048
> max_thread_proc 256 256
>
>
> Adam.

Glance plus seems promising but only lists a few theads by default.
Where is decent doco? Man page is incomplete so is online help.

Adam.

Adam Skeggs <fun@optusnet.com.au> writes:

> A process has thrown sig 11 on a pthread_create but I don't belive at
> the moment that the process itself blew it's thread limit (256).

Even if it did run out of some resource, you'd get an error from
pthread_create() -- ENOMEM or some such. Unix does not usually
signal resource exhaustion via "trowing SIGSEGV".

You should look for a bug in your program instead.

Cheers,
--
In order to understand recursion you must first understand recursion.
Remove /-nsp/ for email.

You might also be overflowing the thread-specific stack, which
defaults on some older versions of HP-UX to something small like 64K
bytes. That causes a SIGSEGV, usually pretty early in running the user
code of the new thread. There's a pthreads API call to set it to
something bigger, though its name escapes me at the moment.

- Carl Burch
HP WDB Debugger Team