Hello,
I'm running Oracle rdbms v7.2.3 (no MTS) under digital Unix 3.2D-1.
I've just noticed that my ps output is somewhat strange when I list oracle
processes (sql*net connections) with 'ps auxww' under digital UNIX 3.2D-1.
First: The VSZ of the processes is quite big (Someone told me that the
attached SGA shared memory segment was included in the VSZ number reported
by ps under digital UNIX, please confirm)
Second: The Resident size (RSZ) varies, but it tends to be quite large
(Since we added 512 Mb to the system, it seems to have somewhat decreased
tough...) and for SOME processes, there is no unit (K's or M's) beside the
figure listed in the ps output. When there is no units, can I assume
KiloBytes ? Is ps under DU bogus or what ?
Third: I would like to know what are the RSZ of other people using a similar
configuration as mine. Please send me your 'ps auxww | grep ora' output so I
can compare.
If anyone can help me explain/reduce the memory requirements of the sql*net
processes, please feel free to send me some suggestions. I've tried tu run
MTS but got in 2 specific problems:
a) My shared pool was too small (even at 20Mb). I have problemes evaluating
the Shared Pool requirement for my application.
b) I had a bunch of the following kind of error messages in my alert log
that I did not like in a production environment (I logged a tar with oracle
about this but got no specific answer yet. I assume there is a bug with MTS
under DU):
Fri Feb 7 10:59:07 1997
Errors in file /m02/oradbs/lpprod/logs/s001_18866.trc:
ORA-00600: internal error code, arguments: [17271], [instantiation space
leak], [], [], [], [], [], []
Thanks !
NOTE: Attached is a ps auxww | grep ora listing.
Guy Dallaire
dallaire_at_total.net
"God only knows if god exists"
Received on Thu Feb 27 1997 - 22:09:42 NZDT