A Performance test was configured to run on 4 different Private Locations and was configured to run for a configured duration of 12 hours.
Approximately 1 hour after the test started, it ended, but there were no errors reported indicating the reason that the test ended prematurely.
Release : SAAS
Component : BLAZEMETER PERFORMANCE TESTING
Filtering the metrics by Private Location revealed that metrics were never collected from 1 of the configured 4 Private Locations. It was discovered that only the BlazeMeter docker crane image was installed on the problematic Private Location. This was determined by executing the following docker command on this Private Location:
sudo docker images
Performance tests, by default, are configured to start when all engines configured in the test are ready. BlazeMeter will wait one hour for all engines to be ready to run before starting the test. This can be by-passed by unselecting the Synchronized start option on the test launch page:
If the Synchronized start option is selected, then a test can be forced to start from the test booting page by selecting the Force Start link once any engines are in a ready state:
BlazeMeter will wait approximately one hour for all of the engines configured in a test to reach the ready state. If any of the configured engines fails to reach a ready state after an hour, then BlazeMeter will stop the test. If a test was forced to start, or if a synchronized start was not configured, then the tests running on the engines that were started will be automatically stopped and a test will terminate before the end of the configured duration.
A message similar to the following will be recorded in the bzt.log files from the engines that were started before all engines were in a ready state:
[2021-07-28 11:35:08,141 INFO Engine.cloud-utils] Got stop command
Assure that all BlazeMeter docker images are installed on the Private Locations configured in the Performance test before starting the test.