max_mem_window / fork errors

[TD Clark]

We are running multiple Sybase engines - they both need at least 2GB of memory. To get around the 2.75GB limitation in HPUX we decided to use memory windows - allowing us to use the max memory possible for each db engine. We set the max_mem_window to 2. Things worked well with Sybase configured with just 2GB of memory for each engine, but when we bumped one of the Sybase engines to 2.5GB we started getting fork errors when users logged in. We have a total of 6.7GB on the System (L-1000 running HPUX 11.0). We are using memory window 1 - the white paper doesn't get into a lot of detail on this - any thoughts?! Until we determine the problem we have bumped the Sybase configuration down to 2GB and all is fin

[Bill Hassell]

The Sybase processes must be only 32 bits (rather than 64 bits where there are no practical limitations). Read the white papers in /usr/share/doc concerning how to modify programs (using chatr) to utilize up to 3 quadrants of 32 bit memory mapping to attain as much as 2.75Gb for a single 32 bit process.

[Sandip Ghosh]

Whenever you are trying to do that try like this way.

Bring down both the sybase instancefirst start the 2.5GB Engine first and then start the 2GB Engine. It should work.

Sandip