Web Server

FreddyFrancis

FreddyFrancis

Well Known Member
I am currently running a JAS server on OAS 10.1.2.0.2

however i find that the CPU utilisation going to 100% with the java.exe using 98-100%.There are around 6 java.exe which got created.

can somebody tell me how to rectify this problem. THe heap memory is utilized only 20.4 %. Also this happens even with few users( 4-5) logged in.

Freddy,
 
Freddy.



We are on 8.95 (going to 8.96) and we are having the exact same
problem. From our early testing we see that 8.96 does not solve the
issue. We have narrowed it down to "Scroll to End" and upload actions by
end users. So we are at a loss. We worked with Oracle and made some
changes to the webclient.war file and it seemed to help somewhat. Our
next move is to figure out if we are having a custom code issue.
Currently we are having to bounce our Web servers in Production every
couple of days.



Anyone with ideas....please let me know!!!!



E811 SP1 AIX/Oracle Linux Web 8.95_F1 going to 8.96.1.3
 
Re: RE: Web Server

That's interesting that you have determined "scroll to end" as part of the issue. With that said, I've had this issue at least once a month. I automatically reboot the web servers everynight and this has kind of cleared up the issue. Just so you know, I'm on 8.96 as well. I've called oracle but they were of no help.
 
Re: RE: Web Server

I remember seeing a post on this not too long ago. Try searching for it, I believe there was some useful information in it.
 
Re: RE: Web Server

it seems the same ( rebooting is reducing the CPU utilization) however i do not find that to be a great solution..

thanks JR, let me chk the LIST thoroughly again
 
Generally speaking you should be rebooting your servers everynight. We are running 8.12 and started out with 8.96_G1 and put in 8.96.1.2 about 1 month ago. We have never experienced any problems but we do reboot nightly. How many JVM's do you have loaded? How many processors and what type (dual core..) How much memory do you have allocated for the JVM's and what is the heap size configured for these JVM's?
 
[ QUOTE ]
Generally speaking you should be rebooting your servers everynight. We are running 8.12 and started out with 8.96_G1 and put in 8.96.1.2 about 1 month ago. We have never experienced any problems but we do reboot nightly. How many JVM's do you have loaded? How many processors and what type (dual core..) How much memory do you have allocated for the JVM's and what is the heap size configured for these JVM's?

[/ QUOTE ]


Huh? Reboot nightly? Now that is just silly.
 
I concur. We are running OAS on AIX and do not reboot more than once per month. Sometimes we go a couple or three months before a reboot. This includes cycling OAS - it coincides with reboots, not something we have to do.
 
I think that Windows is showing its inability to handle memory leaks properly - compared to the other platforms. You might have to indeed reset a windows JAS server more often than other platforms. I have also noticed this.
 
[ QUOTE ]
I think that Windows is showing its inability to handle memory leaks properly - compared to the other platforms. You might have to indeed reset a windows JAS server more often than other platforms. I have also noticed this.

[/ QUOTE ]


I just got done restarting an entirely Windows-based system last night for the installation of BackupExec and ruined my 3 month uptime. Now I understand that, had this been Unix, I likely would not have to to restart after installing the agent but, well I guess that is a Windows weakness.


Signed,
MS Fanboi
 
I agree with you. Unless you have hardware sitting on the side (for a
24x7 shop) rebooting nightly is out of the question. The "rebooting
nightly" is the old standard JDE help desk line motto. Kind of like:
"Send us the logs and we will look at it".





AIXSUX
 
[ QUOTE ]
I am currently running a JAS server on OAS 10.1.2.0.2

however i find that the CPU utilisation going to 100% with the java.exe using 98-100%.There are around 6 java.exe which got created.

can somebody tell me how to rectify this problem. THe heap memory is utilized only 20.4 %. Also this happens even with few users( 4-5) logged in.

Freddy,

[/ QUOTE ]

We are having this same issue (WS2k3, OAS 10.1.2.0.2) and we're still in development, so there are very few users using the system. It will run fune from anywhere between 3 days to a week but the CPU will eventually jump to 100% on java.exe - the one associated with JasOC4J's default_island. Looking at the MANY threads (guessing about 100) associated with this particular process, just 2 were using the bulk of the CPU. Beyond this, I'm not sure what to look at exactly or what would cause this problem, so if anyone could suggest something to look at or try, I'm very interested.

The commandline for the java.exe that's going bad is:

"E:\OAS\jdk\bin\java.exe" "-server" "-Xrs" "-Djava.security.policy=E:\OAS\j2ee\JasOC4J\config\java2.policy" "-Djava.awt.headless=true" "-Xms1024m" "-Xmx1024m" "-Doracle.ons.oraclehome=E:\OAS" "-Doracle.home=E:\OAS" "-Doracle.ons.oracleconfighome=E:\OAS" "-Doracle.ons.clustername=C_123.123.123.123.910040.113681438d4.-8000" "-Doracle.ons.instancename=OASDV.cosrvap62.rtc.ts.local" "-Dopmn.compatible=904" "-Doracle.ons.indexid=JasOC4J.default_island.1" "-Doracle.ons.uid=1375993944" "-Doracle.oc4j.instancename=JasOC4J" "-Doracle.oc4j.islandname=default_island" "-DOPMN=true" "-jar" "oc4j.jar" "-config" "E:\OAS\j2ee\JasOC4J\config\server.xml" "-properties" "-properties" "-ports" "ajp:12501,rmi:12401,jms:12601"

You will notice that the contractor that is responsible for installing the system upped the heap to 1024 from the default 512, this was to try and fix the issue. It did not.

I also agree that rebooting nightly is unacceptable. We're a 24/7 shop.

EDIT: Well it just went to 100% only 11 hours after it was last restarted.

One thing I noticed is that the latest JDBC driver wasn't installed per p.20 of the HTML Web Server Installation Guide. Do you think that would have anything to do with it? I'll be installing it tonight and will report back either way.
 
Just a followup to my last message, the JDBC driver upgrade didn't do a thing.
confused.gif
 
Breakthrough! I ran a file monitor against the java.exe process that was going at 100% and it seems that all that time is being spent writing to two log files:

\OAS\j2ee\JasOC4J\applications\jasapp\webclient\WEB-INF\classes\JdeRoot_20070914.log
and
\OAS\j2ee\JasOC4J\log\jderoot_20070914.log

Checking out these files, I found the same error OVER and OVER and OVER and OVER again:

14 Sep 2007 15:28:23,384 [SEVERE] - [TE] isLegalParent: Couldn't find task id: 00142522600000-GHQJYURUTR-AWTWVTOOXY in F9000 table but task is in F9001 as a parent

Looking up this particular system-id, it is associated with the conversion (based on the date). Sure enough, it's in F9001 but not in F9000.

I will search for similar errors in other threads and post back if I find anything.

In the meantime, if I haven't posted back yet, and you know what this is caused by or how it can be fixed, please post here. Thanks!
 
Back
Top