Re: HTTP Interface gives segmentation fault

[Jeetendra Singh]

I couldn't find a Softice like tool to work on Linux. I tried gdb but its no good for multi threaded environment.   The solution you proposed may work for once a week crash but in my case it is sometimes just 500 messages or it has gone upto 50K at times when the crash takes place. With a rate in excess of 20 msgs/sec (on my test setup where I am load testing it right now), I would not prefer this solution.   Can somebody throw some light on this crash?   I am including devel people in it. May be they need to know and can help.   TIA, Jeetendra ----- Original Message ----- From: Navjot Singh To: Jeetendra Singh ; 'Kannel Users' Sent: Thursday, September 04, 2003 4:21 PM Subject: RE: HTTP Interface gives segmentation fault I had also observed this problem around 2 months back but little differently.   I was using HTTP SMSC relay and figured out that after some heavy traffic, the kannel just keeps mum. My OS(Linux 7.3)  doesn't show me Segmentation fault but it just has to be restarted to bring back to it's normal life.   I fixed the problem by listening to kannel/status at fixed periods. kill the earlier PID and starts fresh. Not a nice solution but my problem is also not frequent. Once in a week or so.   Curious to know why this happens? If you have some nice debugger(like softice) / memory snapshot tool, just try to run and send bulk SMS to force the kannel to throw Segmentation fault. You will get to know what the problem is. Unless someone can come up with a known reason.   regards Navjot Singh -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]On Behalf Of Jeetendra Singh Sent: Thursday, September 04, 2003 3:55 PM To: 'Kannel Users' Subject: HTTP Interface gives segmentation fault I am facing a very peculiar problem. Am using Kannel to interface with a SMSC which gives HTTP interface to send sms. It has been observed that this Kannel gateway is dying due to segmentation fault after handling some random number of messages . We have tried it with varied throughput but it dies always. The number of messages handled has varied from 500-50K. The reason for seg fault is beyond me.   Has anybody else observed the same problem?   I am using 18th August Kannel Snapshot on Linux Red Hat OS.   TIA, Jeetendra