WebSphere Hung Threads

Hung Threads on WebSphere Application Server WebSphere Support Team technical resolutions to crashes, troubleshooting, and known issues articles.

Support Help

Hangs and Hung Thread Messages

Hung threads, also called “hangs,” are threads that have been executing the same request for longer than expected. In WebSphere, the Hung Thread Detector will report threads that “may be hung”, once the request has been running longer than the specified threshold (10 minutes by default). These messages are logged in the server’s SystemOut.log, and include the thread’s ID, its call stack, and the number of total threads on the server that may be hung.

Hung thread messages do not always indicate a problem, as some requests may be expected to run for over 10 minutes (e.g. long-running queries, large reports, etc.). For this reason, it is important to know a thread’s purpose when attempting to diagnose a hang. In some cases, threads reported as hung will eventually continue, and such actions, along with the total length of time the thread was active, are also logged to the SystemOut.log.

In other cases, hung threads may never complete. This may be due to the thread waiting on a response from a backend server or database, a bad network connection, or a defect in the application code that is preventing threads from continuing. Given enough time, the number of hung threads can increase until the entire thread pool is hung, which may cause the app server to become completely unresponsive.

Slow Performance

At times, an application may not be performing as quickly as it once did or as quickly as expected. This may be because a section of application code is running slower, a bottleneck is causing multiple threads to compete for the same resource, or a backend server is not responding in a timely manner. Suboptimal tuning of garbage collection can also cause performance impacts. If threads are running long than expected, but still shorter than the WebSphere Hung Thread Detector threshold, these performance issues will not get reported in the server logs, so diagnosis may require multiple rounds of data captured during the performance issue. If the performance impact is very small (i.e. less than 3 seconds), Request Metrics (for WebSphere 8.5 and 9.0) and Request Timing (for WebSphere Liberty) can be useful to help diagnose the issue.

Short Performance Delays

MustGather: Short performance delays in WebSphere Application Server

Getting performance data from request metrics

Request timing: Diagnosing slow requests on WAS Liberty

MustGathers

MustGather: Performance, hang, or high CPU issues with WebSphere Application Server on Linux

MustGather: Performance, hang, or high CPU issues with WebSphere Application Server on AIX

MustGather: Performance, Hang, or High CPU Issues on Windows

MustGather: Performance, hang, or high CPU issues with WebSphere Application Server on Linux on Containers

Hang Detection Policy

Configuring the WebSphere Application Server hung thread detector to produce javacores

Configuring the hang detection policy

Known Hang Issues

Multithreaded Applications Core Dump or Hang Applies to AIX 7200-05

“IBM prides itself on delivering world class software support with highly skilled, customer-focused people. ”

Return to 101 home

Contact Support