I have a Domino 6.02 CF2 server set up on Redhat 7.2. We are testing it using the server.load utility. When running the server.load R5 simple mail routing test and simulating a load of 500 users the server will run for anywhere between 1 and 4 hours before crashing with the following error on the console: "PANIC: Cannot Attach to Shared Memory Region De to Insufficient Access"This exact error is described by Lotus here: http://www-1.ibm.com/support/docview.wss?rs=0&q=libnotes.so&uid=swg21114726&loc=en_US&cs=utf-8&cc=us&lang=en
After confirming that we are pointing to the correct libnotes.so.0 as described in the technote, we still have this problem consistently. Does anybody have any experience with this error?
Actually, I did see that entry… a Lotus support person on crashes downplayed that technote in my circumstances for some reason.Evidently there have been a few companies experiencing such issues regarding message tracking and another task (index updating perhaps) where there are memory conflicts such writing over the same memory space. The temporary fix is to wipe out the mtdata subdirectory. From my understanding, Lotus has not yet been able to determine exactly what part of the existing code is causing the conflict.
What are the hardware limits of the box (memory / processor, etc)? You have my curiousity because a client and myself are running on a RED HAT SYSTEM. Though we are using 7.3 – I know it is not certified, but it does work.
I am wondering if this problem is just specific to 7.2.
Subject: RE: Domino on linux consistently crashing
This is on a P4 2.4 ghz with 4 gigs of ram. An IBM x335 server. The OS is actually RedHat Advanced server, which is just a variant of 7.2. We have other domino servers running under various flavors of linux, but none with the load we plan to put on this one, and this one runs fine otherwise, but crashes under the load created by server.load. Have you had a heavy load on yours?
Subject: RE: Domino on linux consistently crashing
The heavy load on mine is not from server.load though. When I have stressed test the box, it was from using users when the box was not classified production. It did not have the fail you are noting, but I was on 7.3 though and no advanace server modules.
Have you testing with real users – to verify that server.load is giving true results? I am not saying it is a bad thing, but I wonder if you are getting a problem because of 500 people using their email to an extreme for that long in time.
In my experience (and the environment I manage have more than 1000 users), I don’t see people using their mail box for over 4 hours in one day. Mail itself is a simple application. Which is why I question the results from server.load. With that hardware you have a box much bigger than what is needed for 500 mail users.
HTH – CHeers-- feel free to email should you have other q’s