Capture logs for slow response

Owned by Karl Henselin
Last updated: Oct 25, 2023 by Divij.Sehgal
2 min read

Problem

If FlexDeploy is responding very slowly, provide the following details to support team

Solution

See details below.

  • Provide exact time of issue.

  • Explain issue in detail including screen shots.

  • Provide output of sar and free -m commands on FlexDeploy server.

  • Increase log level to FINEST for two loggers - flexagon.fd.model2.jdbc.framework.proxies.FlexOraclePreparedStatementProxy and flexagon.fd.model2.jdbc.framework.impl.FlexCRUDStatement.

    • Look for SLOW, SLOWER, SLOWEST word in log files.

  • Capture server logs including .out files, flexdeploy*log files, gc logs as shown below.

  • Go to Menu → Admin Operations

    Select Change Log Level Operation and Change Log Level to FINEST

    Select View and Download Logs Operation and Download Last 24 Hour Logs

    After Downloading, Make sure to Change Log Level to INFO

  • Capture thread dumps and include it with other details.

# First find process id by using ps command, for example ps -ef | grep java # Locate JAVA_HOME and cd to execute jstack cd $JAVA_HOME/bin # Take 5 thread dumps at least 10 seconds apart. See below, where # will be 1, 2, 3, 4, 5 ./jstack -l <pid> > /var/tmp/threaddump_#.txt # if jstack does not work, use kill -3 <pid> to get thread dumps.

The following macros are not currently supported in the footer:
  • style