I have this weird something and couldn't reach to a reasonable explanation.
First of all, I run OpenACS 5 (W .LRN) Over RH AS2.1/Oracle184.108.40.206
I have another back-end of Oracle Applications connected to the Same Web Database (A fertile environment for lock conflict)
Anyhow, Sometimes the Web Server hangs vaguely (for a while only) then it returns normal.
I suspected that a lock somehow (from the other back-end) is preventing the server from doing something... (Bu, why the whole server!!)
Here are some notices:
The server log when in hang status
[23/Mar/2004:07:51:35][16469.1228820][-sched:idle3-] Notice: starting
[23/Mar/2004:07:51:36][16469.1245205][-sched:idle4-] Notice: starting
[23/Mar/2004:07:51:37][16469.1261590][-sched:idle5-] Notice: starting
[23/Mar/2004:07:52:50][16469.1277975][-sched:idle6-] Notice: starting
[23/Mar/2004:08:27:50][16469.32771][-sched-] Warning: sched: excessive time taken by proc 6 (9 seconds)
[23/Mar/2004:08:28:01][16469.32771][-sched-] Warning: sched: excessive time taken by proc 10 (9 seconds)
I greped the whole acs directory and found no procedure who echoes this "starting" alone
I've monitored the locks while the server hanging and found that some locks are places on some tables used by the web user.
And the normal return of the web server comes after those locks end.
So it is obvious somehow that these locks are the primary accused reason??
My other notice (and I dont know if it is related by anyhow) is that the OpenACS database user places a lock on BULK_MAIL_MESSAGES frequently. (and sometimes this lock stays for ever)
Can anybody light the thing up??