[omniORB] Problems with parallel connections under Linux
Attila.Pletyak@anemo.com
Attila.Pletyak@anemo.com
Thu, 3 May 2001 10:28:26 +0200
One problem can be that the concurrent threads try to work with a shared
resource without thread synchronizations. I used to have segmentation
faults in that case. If there are shared resources, try to use a mutex or
rwlock.
Suse's gcc is good for omniORB purposes since (at least) 6.3 (that's what
we tried).
There is a memory profiling tool I sometimes use, called MemProf. This is
developped by a Red Hat team member, Owen Taylor.
http://www.redhat.com/otaylor/memprof/index.html
Sincerely,
Attila Pletyak
Anemo Ltd.
Hungary
mailto:attila.pletyak@anemo.com
http://www.anemo.com/
"Hymmen, Holger"
<holger.hymmen@comline.de> To: "'omniorb-list@uk.research.att.com'" <omniorb-list@uk.research.att.com>
Sent by: cc:
owner-omniorb-list@uk.resear Subject: [omniORB] Problems with parallel connections under Linux
ch.att.com
05/03/01 10:04 AM
I recently ported an omniORB 3.0.2 based server to Linux (Suse 7.0) which
has been developed by others under Windows NT. When I tested the server
with
a single connection everything was fine, but when I opened a second
connection the server crashed. Right now, I am pretty clueless as to where
the problem stems from, because it crashes at different points most of the
time. I was told that the server can handle at least 4 connections before
it
starts having problems under Windows. I am not sure if the gcc 2.95.2 that
came with Suse 7.0 was compiled with --enable-threads or not, if it is a
problem with the Informix database under Linux or simply a memory leak
problem in the code. Has anyone experienced problems with Suse or Informix?
Does anyone know a free program like purify, with which I can look for
memory leaks under Linux?
Thanks in advance
Holger