This day we met the following problem:
- Oracle 8.1.7.3 Enterprise Edition (SGA 18 Gb)
Sun Fire 4800 Solaris 8 32 Gb RAM
- At some moment we saw there wasn't enough memory in the server
(TOP shown less than 50 Mb).
- We found the process for which SIZE and RES fields of TOP output
are equal 18 Gb.
- It was a user process: ps -ef | grep shown the output like this
one: oraclework (LOCAL=NO)
- But we didn't see any information about this process (SPID field)
in V$PROCESS view. Aand the starting time for this process (output of
PS command) was earlier than the time of last DB restarting.
- We killed this process (kill -9)
- After it we had a lot free memory in the system. But it became
impossible to connect DB (sqlplus internal, svrmgrl internal) because
of the sessions hanging.
- There isn't any information about it in the alert.log or trace
files.
- After restarting the server we resolved this problem.
Any comments, please.
- about this problem and its solution;
- or any further ways to investigate it if it happens again.
Thanks in advance.
Received on Thu May 25 2006 - 04:51:07 CDT