Mesos can run multiple tasks inside a single executor, but this can be confusing because the executor ID remains the same across multiple tasks on a single host. The scheduler is responsible for setting a unique task name / ID for identifying a process run by a shared executor.
Currently, this plugin can only capture metrics on a per-executor basis, adding an executor_id tag to the measurement. This may have (not sure, needs tests) the unintended side effect of only reporting resource utilization for one (but not more than one) task running inside a given executor.
This enhancement is to capture all running tasks for a given executor (which will be >=1) and add a new tag named task_id. This will ensure this plugin is shipping all available metrics to the Snap processor and publisher, and allow the user to group measurements by executor_idor identify individual tasks by task_id in their time series database of choice.
Continued from #30.
Mesos can run multiple tasks inside a single executor, but this can be confusing because the executor ID remains the same across multiple tasks on a single host. The scheduler is responsible for setting a unique task name / ID for identifying a process run by a shared executor.
Currently, this plugin can only capture metrics on a per-executor basis, adding an
executor_idtag to the measurement. This may have (not sure, needs tests) the unintended side effect of only reporting resource utilization for one (but not more than one) task running inside a given executor.This enhancement is to capture all running tasks for a given executor (which will be
>=1) and add a new tag namedtask_id. This will ensure this plugin is shipping all available metrics to the Snap processor and publisher, and allow the user to group measurements byexecutor_idor identify individual tasks bytask_idin their time series database of choice.cc @lanejlanej