Re: configuring monit's httpd

[Christian Hopp]

On Wed, 14 Jan 2004, Allen Shaw wrote:

Hi!

> (...) I get gdb output like this:
>
> #0  0x0805c959 in fillprocesstree (pt=0x80847e0) at process/common.c:263
> 263     fillprocesstree(pt->children[i]);

Oh, seems to be my domain. )-:

> (gdb) bt
> #0  0x0805c959 in fillprocesstree (pt=0x80847e0) at process/common.c:263
> #1  0x0805c95e in fillprocesstree (pt=0x80847e0) at process/common.c:263
> #2  0x0805c95e in fillprocesstree (pt=0x80847e0) at process/common.c:263
> #3  0x0805c95e in fillprocesstree (pt=0x80847e0) at process/common.c:263
> ..
> #261926 0x0805c95e in fillprocesstree (pt=0x80847e0) at process/common.c:263
> #261927 0x0805c95e in fillprocesstree (pt=0x80847e0) at process/common.c:263
> #261928 0x080503c8 in initprocesstree (reference=0xbfffe548)
>     at monit_process.c:226
> #261929 0x08055ead in validate () at validate.c:149
> #261930 0x0804fe25 in do_default () at monitor.c:562
> #261931 0x0804f97e in do_action (args=0xbfffe604) at monitor.c:348
> #261932 0x0804f649 in main (argc=2, argv=0xbfffe604) at monitor.c:115
> #261933 0x42017589 in __libc_start_main () from /lib/i686/libc.so.6
> (gdb) quit

Mhh... I fear there is a loop in the process tree.  But how could this
happen.... and why reproducible?  May be some fragmented proc info or
race conditions while reading the proc info (new processes, changed
ppids).

Okay... I have added a ring move protection in the fillprocesstree
code.  That means that every entity can be visited only once.  In case
an entity is visited again it is treated like a leaf and the recursion
ends.

I have checked in the new code to CVS.

C.Hopp

-- 
Christian Hopp                                email: address@hidden
Institut für Elektrische Informationstechnik             fon: +49-5323-72-2113
TU Clausthal, Leibnizstr. 28, 38678 Clausthal-Zellerf.   fax: +49-5323-72-3197
                             pgpkey: https://www.iei.tu-clausthal.de/pgp-keys/