Skip to end of metadata
Go to start of metadata

In case of hanging issues, when the portal or any applications on AS Java do not response we need to define where the resource bottleneck happens between client machine and application server.

From user point of view following cases can be differentiated:

 

 The portal does not responsie, the hourglass animation is running continuously at a specific navigation step

Useful logs and traces:

  • Client side HTTP traffice captured by the free version of HTTPWatch tool. (www.httpwatch.com)
  • On server side 3-4 thread dumps taken in 60 seconds interval during the hanging occurs.
    #1095473 - How to get a full thread dump in AS Java
  • Logs from work folder - /usr/sap/<SID>/<inst_nr>/work
    dev_server*, std_server*
  • Latest defaulttrace files from Java server nodes
    /usr/sap/<SID>/<inst_nr>/j2ee/cluster/serverX/log/defaulttrace

 

 The portal navigation terminates with a connection timeout to ICM process

Useful logs and traces

  • Client side HTTP traffice captured by the free version of HTTPWatch tool. (www.httpwatch.com)
  • On server side 3-4 thread dumps taken in 60 seconds interval during the hanging occurs.
    #1095473 - How to get a full thread dump in AS Java
  • Logs from work folder - /usr/sap/<SID>/<inst_nr>/work
    dev_server*, std_server*, dev_icm*
  • Latest defaulttrace files from Java server nodes
    /usr/sap/<SID>/<inst_nr>/j2ee/cluster/serverX/log/defaulttrace

 

 The browser does not response to any actions on portal

Useful logs and traces

  • Client side HTTP traffic
  • Thread dumps
    #1095473 - How to get a full thread dump in AS Java
  • Logs from work folder - /usr/sap/<SID>/<inst_nr>/work
    dev_server*, std_server*, dev_icm*
  • Latest defaulttrace files from Java server nodes
    /usr/sap/<SID>/<inst_nr>/j2ee/cluster/serverX/log/defaulttrace

 

 The portal navigation is very slow

Useful logs and traces

  • Client side HTTP traffic
  • Thread dumps
    #1095473 - How to get a full thread dump in AS Java
  • Logs from work folder - /usr/sap/<SID>/<inst_nr>/work
    dev_server*, std_server*, dev_icm*
  • Latest defaulttrace files from Java server nodes
    /usr/sap/<SID>/<inst_nr>/j2ee/cluster/serverX/log/defaulttrace
  • SAPJVM profiler
    #1995883 - Analyzing slow AS-JAVA startup using the SAPJVM profiler

 

 The logon to the portal takes very long time

Useful logs and traces

  • Client side HTTP traffic
  • Trace with security troubleshooting wizard, "Authentication" incident. Report is stored in a ZIP archive.
    #1332726 - Troubleshooting Wizard
  • Thread dumps
    #1095473 - How to get a full thread dump in AS Java
  • Logs from work folder - /usr/sap/<SID>/<inst_nr>/work
    dev_server*, std_server*, dev_icm*
  • Latest defaulttrace files from Java server nodes
    /usr/sap/<SID>/<inst_nr>/j2ee/cluster/serverX/log/defaulttrace
  • SAPJVM profiler
    #1995883 - Analyzing slow AS-JAVA startup using the SAPJVM profiler

 

 XI messages are not processed

Useful logs and traces

  • XPI Inspector trace
    #1514898 - XPI Inspector for troubleshooting XI
  • Thread dumps
    #1095473 - How to get a full thread dump in AS Java
  • Logs from work folder - /usr/sap/<SID>/<inst_nr>/work
    dev_server*, std_server*, dev_icm*

 

 Hanging deployment
Remarks on the tools:

To capture the client side HTTP traffic use the free version of HTTPWatch tool. The tool is available on www.httpwatch.com size. The result can be exported into a .HWL file.

The thread dumps can reveal the status of the AS Java server nodes. It is important to trigger 3-4 thread dumps in 60 seconds interval during the hanging issue happens. The thread dumpl will show the long running and blocking threads. The output of the thread dump is stored in the dev_serverX and std_serverX.out files in the work folder (/usr/sap/<SID>/<inst_nr>/work).
#1095473 - How to get a full thread dump in AS Java

MMC snapshot about the whole system. The snapshot contains information about thread usage, status of server nodes, ICM processes.
#1847251 - How to create an MMC snapshot about an SAP system

SAPJVM profiler. The profiler collects statistics about the status of the stacks in 20ms intervals during the profiler is active. The profile can show the most CPU consuming methods, threads.
#1995883 - Analyzing slow AS-JAVA startup using the SAPJVM profiler

 

  • No labels