I have/had a site that occasionally locks up and is core dumping with
the 'waited too long for row cache enqueue lock'. (they had to shutdown
abort in 7.1.4) and so far have been unable to find out (thru books and
some interface with Oracle corp )that, IF it's a parameter problem, what
to make bigger. Yesterday I suggested they increase database block
buffers by 20% (to 10,000), and enqueue resources from 1220 to 1720.
Redo logs weren't switching over (presumably because they weren't doing
anything). utllockt wouldn't run/finish. We couldn't even look at
almost anything in the database because queries, etc would not finish.
The machine has more memory than we are currently using and the SGA is
about 140M. Solaris 2.3. There were 'a lot' of users logged on and the
appls typically reference a lot of tables, I understand. (definitely not
an OLTP system)
Would appreciate any ideas of what I should be looking for or where to
go for more info. I have looked at Oracle Press's backup and recovery,
Oracle's DBA book (7) and the Orange Tuning book (not an Oracle press
book)
thanks
Joan
Received on Thu Apr 23 1998 - 10:38:21 CDT
