Handling WorkZone agent errors

Errors can occur during the execution of WorkZone agent jobs. These errors can cause failure or unforeseen behavior of the agent, and affect WorkZone functionality and performance. You can fix some of the errors or restart/reschedule agent jobs to alleviate some of the effects.

The following list explains some of the most common errors and how to fix them. Please note that the list is in no way exhaustive and does not include all potential agent errors.

AgentFIX

No document to index

This error often occurs if the user has canceled the creation of a document. A row will typically appear in the SERVICE_QUEUE table indicating an AgentFIX job is to be scheduled, but there is no document to index.

To fix this error, delete the row from the SERVICE_QUEUE table with the following SQL command:

DELETE FROM SERVICE_QUEUE WHERE KEY=<####> AND AGENT_TYPE=’FIX’

<####> is the job identifier from the alert in the event log.

Indexing an unsupported document

If the AgentFIX agent attempts to index an unsupported document, an error will be triggered. Unsupported documents are documents that are not supported by the INSO viewer.

To fix this error, delete the row from the SERVICE_QUEUE table with the following SQL command:

DELETE FROM SERVICE_QUEUE WHERE KEY=<####> AND AGENT_TYPE=’FIX’

<####> is the job identifier from the alert in the event log.

Other errors

Occasionally, errors occur for other reasons which cannot be readily explained or adequately diagnosed. Often, these error can be alleviated by rerunning the AgentFIX job for all failed jobs with the following SQL command:

UPDATE SERVICE_QUEUE SET STATUS=NULL WHERE AGENTTYPE=’FIX’ AND STATUS=2

AgentSUB

Errors do not often occur in AgentSUB jobs due to the nature of the AgentSUB job. Usually AgentSUB jobs fail because of errors in the set up of the subscription itself. Subscription setup errors can be fixed in WorkZone Client and cannot be alleviated by using the AgentSUB.

You can, however, reset the Agent SUB job before checking the subscription setup in WorkZone Client with the following SQL command:

UPDATE SERVICE_QUEUE SET STATUS=NULL WHERE KEY=<####>

<####> is the job identifier from the alert in the event log.

AgentOCR

Stalled OCR job

The most common AgentOCR error occurs when a document cannot be read or processed by the AgentOCR, usually because the document is an image or photograph. Some of these documents will cause the AgentOCR to crash, leaving a row in the SERVICE_QUEUE which is not registered as failed because the job still is considered to be in progress even though it is stalled.

While a failed AgentOCR job can be found by the Failed status (STATUS = 2), a stalled AgentOCR job is more difficult to locate. Usually, the AgentOCR job with the lowest KEY value that hasn't failed is a good candidate.

  1. Locate the AgentOCR job row in the SERVICE_QUEUE table with the lowest KEY value and with a STATUS not equal to 2 (The job hasn't failed yet).
  2. Change the status of the job by using this SQL command:
  3. UPDATE SERVICE_QUEUE SET STATUS=2 WHERE KEY=<####>
    <####> is the lowest KEY value found in step 1.
  4. In the Windows Services Management Console, locate and restart the AgentOCR service.

Process services

Emails are handled by the mail agent which uses the service queue to pick up jobs.



Emails are not received

If an email is not received by an actor, you can use the following command to examine the service queue for mails that failed to generate:

select * from service_queue where agent_type = ‘WZP’ and status = ‘2’;

status = 2 indicates that the mail generation failed and the column status_txt may reveal the cause of the failure. 

Run the job again

Clear the following status to run the job again when the problem is resolved:

update service_queue set status = ‘’ where key = <key number>;