In environments with unreliable network connections such as the Internet, unforeseen exceptions can happen. If the controller loses connection to an agent during a load test or while collecting results, Silk Performer will try to re-establish the connection. However, this might not always be possible and would lead to an incomplete results set.
For such a case, Silk Performer provides a recovery workflow. It allows you to recover results and build a new set of valid data. Recovering results is not an ideal solution, but it prevents you from entirely losing the already gathered data.
The recovery workflow varies, depending on the type of load test you have executed.
Assumed, you are executing a CloudBurst load test. For an unknown reason, the Workbench becomes unresponsive and finally crashes. When you restart the Workbench, the recovery workflow dialog automatically displays. Follow the recovery workflow and review the following fields, which are automatically prefilled:
For CloudBurst load tests, these fields are automatically prefilled. Review the directories and adjust them if necessary
For recovering an on-premise load test, we assume you are executing a load test with agents in your server lab. For some reason the connection to several agents gets lost and cannot be re-established. Thus, the load test ends with an incomplete results set. In this case, click
.Follow the recovery workflow and specify all required directories. The directories are the same as described above. However, the fields are not automatically prefilled. Before you specify a directory for the missing results, make sure to create this directory on the controller and to create a subfolder for every lost agent. Then, copy the results from the agents into the subfolders. Specify all three directories and start the recovery workflow.
Silk Performer will execute several steps to recover the results. If an issue occurs during recovery, the process stops and you can view a log file for more information. Depending on the issue, you can ignore it and let the recovery process continue, or you can restart the process.
Once the recovery is completed, the new set of results displays in the Results tree.