[nos-bbs] anyone else getting JNOS 2 hangs after 24 hours ?

Barry Siegfried k2mf at k2mf.ampr.org
Wed Nov 10 03:12:02 EST 2004


[maiko at pcs.mb.ca wrote]:

> A bit of a set back (darn it all, just when I was on a role).

Or roll?  :)

> I've had one person so far say that after 24 hours, their JNOS 2
> suddenly refuses connections, and takes up major CPU time.  To make
> matters worse, I too have just found it happening on my test system
> here, and I'm hardly doing anything as far as traffic.
>
> so beware ... I need to find out why !!!  I suspect it's related
> to my axudp merge (no doubt I forgot something).
>
> Rats !!!

I doubt that it was due to that.  There is nothing in the AXUDP (or
AXIP into which you merged it) code for that matter that uses any
additional dynamic resources whatsoever.  By dynamic resources, I
mean the dynamic allocation of memory on the fly and/or the use of
additional CPU time in a new process (since there isn't one).

Your problem is more likely in something else you recently did which
*does* dynamically allocate memory and/or creates an additional
process.  Look there *first*.  You can narrow this down by making
a list of your recent changes and looking at anything you did which
dynamically allocates new memory (perhaps there is not a corresponding
free() call) or a new process doesn't get cleared out completely when
it is done and is left being polled on a list through pwait().

Try and look at the *state* of the machine when the problem occurs.
Is there anything left hanging in a (TCP, Net/Rom or AX.25) connection
control block, socket or application process after an operation is
done?  Also, what is the memory state?  Has some additional memory
perhaps been dynamically allocated and then abandoned forever?  These
things will give you clues as to area of code where the problem may
be, and most probably it *is* in an area of code on which you recently
worked.

Also, try turning *off* something that was recently added which uses
new memory and/or process allocating code and see if the problem goes
away.  Finding that spot will also help point you in the right direction.

73, de Barry, K2MF >>
           o
          <|>      Barry Siegfried
+---------/-\---------------------------+
| Internet | bgs at intac.com              |
| HomePage | http://www.intac.com/~bgs  |
+----------+----------------------------+
| Amprnet  | k2mf at k2mf.ampr.org         |
| PBBS     | k2mf at k2ge.#cnj.nj.usa.noam |
+----------+----------------------------+




More information about the nos-bbs mailing list