I used to run Oracle 9i on RedHat 8. The machine was
DELL PowerEdge 16000SC, with Xeon processor, 2 CPUs
and 2GB Ram. The problem I was having with it was that
the database seemed to allow on limited number of active
threads. Whenever it reaches 53 active threads the database
would freeze up and no read/write operation could be performed
and no new user can login. I tried killing some unimportant
threads and as a result a few new users can login. Eventually
I had to reset the database.
But the strange thing is the alert log did not report any error
like nothing is wrong.
Not long after that, I migrated the db to a more powerful machine,
IBM X-series with 4 CPUs still running on Xeon processors, but this
time I use RedHat Advance Server 2.1.
However the problem still persists although with higher limit. The
db would hang when it reaches 63 active processes and again
no error messages in the alert log indicating anything is wrong.
To this date I am still facing this problem. I have to write a script
to automatically reset the database when the problem happens. The db
is used to cater for about 300 users and needs to be up 24 by 7.
I did not have this problem when the db was running on Solaris 8.0.5
before this. But the old machine had hardware failure and I had to migrate
the db to a cheaper hardware, thus the open source solution.
The management is going to acquire even more powerful IBM Xseries machine
next month or so and insisted that we stay with the open source solution.
If anybody out there has any advice, please share.