FOLLOW UP: Increasing SGA size

Just wanted to say thanks to the NET my problem was solved.

Oracle didn't even call me back until 10 days after I originally logged the problem report (TAR) and that was 3 days after I solved the problem with help from the net. Gee, that Oracle support we pay for is really useful. NOT!

Anyway, my problem was that while all I needed to do was increase part of the SGA (the shared pool which is controlled by a shared pool parameter in my INIT.ORA file). Previously I had been increasing all of the parameters in the INIT.ORA file to those for a medium sized system including the number of processes from 50 to 100. This change prevented Oracle from starting up because is attempted to allocate more semaphores than were available.

I had thought we had a sufficient number of semaphores but it turns out UNIX (or is it just ULTRIX) allocates semaphores in blocks so that my one other process that allocates 1 semaphore actually uses up a whole block of 25. Once I realized that, I could see why ORACLE was having a problem with semaphores. In the short term, I simply left the number of processes at 50 (the SMALL size in the INIT.ORA file) and only increased the shared pool size (to the MEDIUM value in the INIT.ORA file). This allowed me to start up ORACLE with an increased SGA (my original problem) without attempting to allocate more semaphores than currently available. When the system manager has more time and when the user load is a little lighter we plan to take the most commonly offered advice....rebuild the kernal with more semaphores (and perhaps a slightly smaller semaphore block size).

                thanks again,
                   jim