Agent and Job Statuses
Agent statuses can be viewed in the Active Agents table under the Jobs tab. Because the Agents manage the scheduling, starting, and stopping of jobs, Agents have their own job-control statuses. The statuses of Agents are aggregated by the Quasar Server as the Agents report their progress. The result is that each job on the Quasar Server will also have a status. This section provides an overview of the Agent and Job statuses.
A typical, successful agent status lifecycle will follow the path in blue.
Agents have their own states, while the server determines the overall job state from the agents’ individual states.
Each of the agent statuses are described in greater detail below.
NOTE: Agents do not hold state and restart the job if there was a reboot. You will have to run the job again.
The Agent has been assigned a job.
The Agent is scheduled to run a job.
The Agent is waiting to enter a window to start the job (for example, if the job window is in the future).
The Agent has entered the time window. The Agent is initialising the job.
It is unlikely that this status will be viewable as it typically completes quickly. However, you may see this state while Quasar is enumerating tables on a large SQL database, for example.
The Agent is inside of the time-window and has started the job process. The Agent is proceeding as intended.
The Agent has failed to start the job, usually because something went wrong during the Agent initialisation process.
The most common causes for this error are trying to scan folders that do not exist or setting an incorrect date / time (e.g. the completion time for the scan is in the past).
The Agent has completed its task and has shut itself down as intended.
The Agent has been shutdown in response to the user aborting the job.
All Agents will be set to USERABORT if the user aborts the job. Individual agents cannot be removed from a job without aborting the entire job.
This may also indicate a full database or a resource being over its limit (e.g., tempspace), causing the job to be aborted.
An Agent may exit the job (without having completed it) for several reasons. The reason for this message is recorded and visible in the Messages view for the job. Some examples of reasons why a job would be exited include:
If the job is running, but the Agent has reached the end-window time without having completed the scan.
If the Agent detects that it has been restarted (for example, due to the computer rebooting) while it is RUNNING the Agent will not complete the job and will be marked as EXITED.
If the Agent shuts down while it is running jobs (for example, because the computer shut down or someone stopped the Quasar Agent Service), all associated jobs will also be shut down.
Job statuses are are viewed in the Jobs table and when you click the name of a job to view more detail. These statuses are controlled either by user actions, or by the server aggregating Agent-statuses.
A typical, successful job status lifecycle will follow the path in blue.
The creation of a new job is in progress.
Jobs are not saved in DRAFT form until at least the first page of the job creation wizard has been completed and the job has been given a name and time window.
A DRAFT job will not be run until the job creation wizard has been completed. Jobs whose status shows as DRAFT can be edited until the wizard is completed.
The job creation wizard has been completed. However, the Quasar Server has not picked the job up.
This status typically happens when the Quasar Server is not running, and an error message will be provided.
The Quasar Server will always pick up SUBMITTED jobs upon startup.
The Quasar Server has received the job and is either ready to push it out to Agents or has already pushed it out to the Agents.
This status will be seen if the job window is in the future.
A job transitions from SCHEDULED to RUNNING when one or more Agents reports that they are RUNNING the associated job.
Job Halt / Completion Statuses
The job halt / completion statuses depend on whether the Agent statuses are uniform.
If/when all agents in a job report uniform statuses, the job status will change immediately. This allows jobs to finish before their end-time window.
The job status for non-uniform Agent statuses are determined by the k_ratio Scan Setting.
If the Agents are not reporting uniformly before the job’s end-time, the job will continue to run until the completion time, based on the rules detailed below.
NOTE: The most common cause for a job not completing is when one of the Agents is not reporting to the Quasar Server. If that happens, the scan will run until the set end-time before reporting a job status.
If all Agents report a status of STARTFAIL, the job status is set to STARTFAIL.
Once the job has reached the set end-time, if there are Agents whose Agent Status is STARTFAIL, the job status will be reported as STARTFAIL.
USERABORT or ABORTED
If all Agents report a status of USERABORT, the job status is set to USERABORT or ABORTED.
EXITED and COMPLETED Statuses
Whether a job is considered to be EXITED or COMPLETED is determined by the determined by the k_ratio Scan Setting.
If all Agents report a status of DONE, the job status is set to COMPLETED.
Once the job has reached the set end-time, if the percentage of Agents that have failed is less than the k_ratio setting, the job will be marked as COMPLETED.
For the purposes of the k_ratio test, SCHED, STARTING, RUNNING, and EXITED Agent statuses are treated the same.
If your job is not completing, check the ‘Inactive Agents’ for more information.
If all Agents report a status of EXITED, the job status is set to EXITED.
If no Agents report back at the end-time of the window, the job status is set to EXITED.
Once the job has reached the set end-time, if the percentage of Agents that have failed is greater than the k_ratio setting, the job will be marked as EXITED.
NOTE: Because data is streamed throughout the scanning process, if an agent is stopped early (e.g., at 99%), all data up to this point will be output to the Quasar database. This means that incomplete jobs may still be useful.
NOTE: If an agent does not show up during the job window, the job’s completion state is not evaluated until the end of the job window.