Troubleshooting
Problem
Content Engine using SQL Server as datasource provider on Windows 2003. Reports high CPU % utilization between 85-100 % on the CE JVM process.
Symptom
During non-peak usage periods on the CE server /Windows 2003 high CPU utilization is seen on CE JVM
Cause
Current issue with SQL JDBC DataSource version 2.0.1803.100. Issue is addressed in Microsoft HF(HotFix) update :
http://support.microsoft.com/kb/977924
Environment
Content Engine 4.5.x
Windows 2003
SQL Server
Diagnosing The Problem
The following actions were used in determining root cause of issue reported:
- Enable verboseGC data. Below are the instructions for doing so in the WebSphere Admin Console:
Enabling verbose garbage collection (verbosegc) in WebSphere Application Server http://www-304.ibm.com/support/docview.wss?uid=swg21114927
- Please follow the instructions for installing TPROF for Windows, a mentioned in the MustGather technote as being our preferred way to collect CPU data.
MustGather: Performance, hang, or high CPU issues on Windows http://www-304.ibm.com/support/docview.wss?uid=swg21111364
- Once the high CPU issue is seen again, recollect the High CPU MustGather data using TPOF and running the windows_hang.py script at the same time.
Once the data has been collected, review the three javacores, TPROF CPU data, and server logs (SystemOut, SystemErr, native_stdout, and native_stderr.log)
- Identification of the current version SQL JBDC driver being used by CE can be view in the CE JVM logs systemout.log
[6/5/12 12:38:31:964 CDT] 00000022 DSConfigurati I DSRA8203I: Database
product name : Microsoft SQL Server
[6/5/12 12:38:31:995 CDT] 00000022 DSConfigurati I DSRA8204I: Database
product version : 9.00.3042
[6/5/12 12:38:31:995 CDT] 00000022 DSConfigurati I DSRA8205I: JDBC
driver name : Microsoft SQL Server JDBC Driver 2.0
[6/5/12 12:38:31:995 CDT] 00000022 DSConfigurati I DSRA8206I: JDBC
driver version : 2.0.1803.100
Resolving The Problem
Review of collected Javacore files showed evidence as follows:
Found 3 threads consuming over 18% each of the cpu. They are perform code from FileNet to persist data to the SQL database and the threads are reading/performing steps to the microsoft/sqlserver:
-
LAB TKS %%% NAMES
PID 61876 88.09 java.exe_0d9c
)) Process_Thread
- LAB TKS %%% NAMES
PID 61876 88.09 java.exe_0d9c
TID 13270 18.89 tid_170c
TID 12985 18.49 tid_1778
TID 12979 18.48 tid_1738
TID 3722 5.30 tid_1034
- Each java/callstack referered to SQL DB calls being issue by FileNet:
Java callstack:
at com/microsoft/sqlserver/jdbc/SQLServerConnection.terminate(
at com/microsoft/sqlserver/jdbc/SQLServerConnection.terminate(
at com/microsoft/sqlserver/jdbc/TDSChannel.read(IOBu
at com/microsoft/sqlserver/jdbc/TDSReader.readPacket(IOBuffe
at com/microsoft/sqlserver/jdbc/TDSReader.nextPacket(IOB
<...com/mircorsoft/sqlserver/jdbc code>
- The above pointed to SQL JDBC as culprit and hence applying the following MS HF fpr SQL JBDC driver update to version 2.0.1803.206 on CE server in question.
===
http://support.microsoft.com/kb/977924
Was this topic helpful?
Document Information
Modified date:
17 June 2018
UID
swg21599191