Issue Summary
This bug occurs in the following scenario:
- A job start running on a remote agent A;
- The remote agent A goes offline;
- This causes the build to fail;
- Re-run the build, which will make the failed job run in another agent;
- During the build, the remote agent A comes back online and reports the job it was building failed;
- The job will be marked as failed.
Steps to Reproduce
The steps below describe how to reproduce this issue in a single machine:
- Create a job and configure a required artifact;
- Create a script task with the following commands:
echo "Testing job failure" > test.txt sleep 60
- Configure a remote agent to connect to Bamboo using a hosts entry;
- Start the build, during the "sleep" remove the Bamboo entry from the file;
- Shutdown the agent and put the hosts entry back;
- Re-run the build;
- While the sleep is running, start the remote agent.
Expected Results
The job keeps running on another agent and the server reports the result from this run.
Actual Results
The server will receive the failed message from the agent that came back online and it will mark the job as failed.
Also, we can see in the build logs the post-build processes messages at the begging of the log:
The below exception is thrown in the xxxxxxx.log file:
simple 13-Oct-2021 17:50:21 Running on server: post build plugin 'NCover Results Collector' simple 13-Oct-2021 17:50:21 Running on server: post build plugin 'Build Hanging Detection Configuration' simple 13-Oct-2021 17:50:21 Running on server: post build plugin 'Clover Delta Calculator' simple 13-Oct-2021 17:50:21 Running on server: post build plugin 'Maven Dependencies Postprocessor' simple 13-Oct-2021 17:50:21 All post build plugins have finished simple 13-Oct-2021 17:50:21 Generating build results summary... simple 13-Oct-2021 17:50:21 Saving build results to disk... simple 13-Oct-2021 17:50:21 Store variable context... simple 13-Oct-2021 17:50:21 Indexing build results... simple 13-Oct-2021 17:50:21 Finished building TEST-AG-JOB1-17. simple 13-Oct-2021 17:50:21 Build TEST - AgentLoop - Default Job #17 (TEST-AG-JOB1-17) started building on agent 192.168.15.5 (2), bamboo version: 8.0.0 simple 13-Oct-2021 17:50:21 Remote agent on host 192.168.15.5 simple 13-Oct-2021 17:50:21 Build working directory is /Users/gluz2/support/agent-home-8.0.0/xml-data/build-dir/TEST-AG-JOB1 simple 13-Oct-2021 17:50:21 Executing build TEST - AgentLoop - Default Job #17 (TEST-AG-JOB1-17) simple 13-Oct-2021 17:50:21 Running pre-build action: VCS Version Collector simple 13-Oct-2021 17:50:21 Starting task 'Script' of type 'com.atlassian.bamboo.plugins.scripttask:task.builder.script' command 13-Oct-2021 17:50:21 Beginning to execute external process for build 'TEST - AgentLoop - Default Job #17 (TEST-AG-JOB1-17)'\n ... running command line: \n (...) Truncated simple 13-Oct-2021 17:50:21 Finished task 'Script' with result: Success simple 13-Oct-2021 17:50:21 Starting task 'Sleep' of type 'com.atlassian.bamboo.plugins.scripttask:task.builder.script' command 13-Oct-2021 17:50:21 Beginning to execute external process for build 'TEST - AgentLoop - Default Job #17 (TEST-AG-JOB1-17)'\n ... running command line: (...) Truncated simple 13-Oct-2021 17:51:21 Finished task 'Sleep' with result: Success simple 13-Oct-2021 17:51:21 Running post build plugin 'Docker Container Cleanup' simple 13-Oct-2021 17:51:21 Running post build plugin 'NCover Results Collector' simple 13-Oct-2021 17:51:21 Running post build plugin 'Clover Results Collector' simple 13-Oct-2021 17:51:21 Running post build plugin 'npm Cache Cleanup' simple 13-Oct-2021 17:51:21 Running post build plugin 'Artifact Copier' simple 13-Oct-2021 17:51:21 Publishing an artifact: Test simple 13-Oct-2021 17:51:21 Finished publishing of artifact Required shared artifact: [Test], pattern: [**/*.txt] in 173.8 ms simple 13-Oct-2021 17:51:21 Finalising the build... simple 13-Oct-2021 17:51:21 Stopping timer. simple 13-Oct-2021 17:51:21 Build TEST-AG-JOB1-17 completed.
Workaround
Currently, there is no known workaround for this behavior. A workaround will be added here when available