If the database server was in performance bottlenecks, there could be various kinds of errors. Here is one case of low memory. First of all, you might see the dmesg from OS-level like this:
...
Oct 11 14:12:43 dbhost cl_runtime: [ID 661778 kern.warning] WARNING: clcomm: memory low: freemem 0x1605
...The above message was detected by the host and it showed the free memory was too low.
In the alert log, similar alert messages generated by PMON were reoccurred like this:
...
Mon Oct 11 15:22:35 2010
ksbsrv: No startup acknowledgement from forked process after 30 seconds
...
Mon Oct 11 15:52:11 2010
ksbsrv: No startup acknowledgement from forked process after 3 seconds
...
Mon Oct 11 16:27:10 2010
ksbsrv: No startup acknowledgement from forked process after 3 seconds
...
Mon Oct 11 16:35:22 2010
ksbsrv: No startup acknowledgement from forked process after 30 seconds
...
Mon Oct 11 16:57:34 2010
ksbsrv: No startup acknowledgement from forked process after 3 seconds
...The root cause was the low memory on the host. At last, we added some physical memory online, such alert messages disappeared.
I think killing some cron jobs could also work, but first, you must identify the top heavy jobs weighting on the host. Here is my another post that might be helpful: How to Monitor Unix Memory Usage by Process