Yes, I logged in DBA. However, the trace files started generating after half an hour. Looks like this site had severe I/O issue. Why I am thinking in that front is ...
1. OS level file copy from one FS to other FS is very slow
2. at least two db writer processes are in the top list (when using unix top)
3. at least one archiver processes is in the top list
4. often check point will be first in the top consuming 80% cpu
5. log writer always complaints that can not allocate new log sequence
6. sequential reads are damn slow ... and so on
This Oracle 10.2.0.4 db is on HP RX 8640 with 24 cores and 48 GB memory. Storage is Hitachi FC SAN.
I am working on to prove with evidence that there an I/O issue.
Any suggestions from your side is much appretiated.
P.S - due to license issues I can not use AWR & ADDM here. I can only work on Statspack.
Try hard to get what you like OR you will be forced to like what you get.
sar or vmstat whatedver u mentioned should be good enough for initial I/O investigation. Then you may get into details using iostat or similar command and at same time get the pid track it what it's doing.