This content has been marked as final. Show 8 replies
You need to provide more information. Specifically
What exactly is 'peak' time ? Do you define that by the number of concurrent requests coming in to the database or some other criterion.
What is the hardware that your APEX installation is running on.
What are the APEX versions and Oracle versions.
Also, the network class waits that you are seeing are in all likelihood 'idle' waits. You are better of obtaining an AWR report for the time when performance has degraded.
There are about 500 clicks per hour during the peak time.
I am using a PC server to host APEX. The memory is 4G.
Oracle version: 188.8.131.52, APEX version: 3.1
The wait time is about 10-30 seconds. The wait event is "virtual circuit wait", the status is "WAITING" or "WAITED KNOWN TIME". I am trying to get AWR report.
I suspect that the switch from apache to EPG might have caused the problem: before we did not have such problems.I'm curious...why did you switch from Apache to the EPG?
Author of Pro Application Express: http://tinyurl.com/3gu7cd
REWARDS: Please remember to mark helpful or correct posts on the forum, not just for my answers but for everyone!
it was in the context of a cross-platform 9.2 to 11.1 migration that
we decided to opt for EPG using the APEX bundle integral to the db
packages (thus, we are still running 3.0.1). I think some documents
mentioned certain advantages of this architecture, but meanwhile I
have my doubts ...
To install APEX, we simply followed Note:457621.1 (How to Configure
Oracle Application Express (APEX) & the Embedded PL/SQL Gateway (EPG)
in an 11G DB).
We currently have a SR open for weeks, were it is tried to track down
the reason for those numerous virtual circuit waits, and found that
all sessions emenate from APEX sessions (with various users and
various apps). Thus, it is quite difficult to track down the reason
for those waits.
Is there an easy procedure to switch to Apache instead of EPG?
in the end support was not able to say which client operation
should have caused those waits. The multithreaded architecture
makes it hard to search deeper. We would had to dig deeper
through network traces ...
APEX supporters recommendation was to switch to the http server
base for APEX:
The http server has better mechanism for traces etc..
(error log, access log, etc..)
I also know that in the next release (4.0) EPG will not be
supported for this reason too.
We will still need to implement the switch back to http server,
but I believe that our problem is not client based and will
vanish with the switch back to http based APEX.
Oracle Database 11g Express Edition Release 184.108.40.206.0 - 64bit Production
SQL> exec dbms_xdb.cfg_update( updatexml( dbms_xdb.cfg_get(), '/xdbconfig/sysconfig/call-timeout/text()', '300', 'xmlns="http://xmlns.oracle.com/xdb/xdbconfig.xsd"'));
BEGIN dbms_xdb.cfg_update( updatexml( dbms_xdb.cfg_get(), '/xdbconfig/sysconfig/call-timeout/text()', '300', 'xmlns="http://xmlns.oracle.com/xdb/xdbconfig.xsd"')); END;
ERROR at line 1:
ORA-06550: line 1, column 28:
PLS-00201: identifier 'UPDATEXML' must be declared
ORA-06550: line 1, column 7:
PL/SQL: Statement ignored
This fails ... please explain ... it was on the post/thread as a possible solution ... now I can't find the rest of the thread where this post / solution was mentioned?