Adaptive Disconnector is a Jenkins plugin designed to help identify potentially unhealthy build agents after a failed build.
Whenever a build fails, the plugin reruns all configured node monitors on the affected agent to check whether any monitored metrics have exceeded their defined thresholds. If a threshold is exceeded, the agent can be disconnected before subsequent jobs are scheduled.
By performing this additional health check at build-failure time, Adaptive Disconnector helps Jenkins react to infrastructure issues and avoid assigning new workloads to agents that may be experiencing problems.
Adaptive Disconnector follows a simple workflow:
- A build running on a Jenkins agent fails.
- The plugin triggers the execution of all configured node monitors on that agent.
- Jenkins evaluates the monitored metrics against their configured thresholds.
- If a monitored threshold is exceeded, the agent can be disconnected.
- Jenkins can then avoid scheduling subsequent jobs on that disconnected agent.
This mechanism provides an additional opportunity to detect infrastructure problems immediately after a build failure.
Adaptive Disconnector can be installed through the Jenkins Plugin Manager.
- Open Manage Jenkins.
- Navigate to Plugins.
- Open the Available plugins tab.
- Search for
Adaptive Disconnector. - Select the plugin and click Install.
You can also find the plugin on the Jenkins Plugin Index:
Adaptive Disconnector – Jenkins Plugin Index
Adaptive Disconnector relies on Jenkins node monitors to evaluate the health of build agents.
The plugin does not introduce a separate set of monitoring thresholds. Instead, it reruns the node monitors already configured in Jenkins when a build fails on an agent.
To use the plugin effectively:
- Configure the appropriate node monitors for your Jenkins environment.
- Define suitable thresholds for the metrics you want Jenkins to monitor.
- Ensure that the monitors are enabled and applicable to the agents you want to evaluate.
When a build fails, the plugin triggers these monitors again, allowing Jenkins to identify agents whose monitored metrics exceed the configured thresholds.
