manifoldcf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Karl Wright (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (CONNECTORS-883) Framework: Job notification thread does not handle notification service interruptions properly
Date Thu, 06 Feb 2014 09:26:10 GMT

    [ https://issues.apache.org/jira/browse/CONNECTORS-883?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13893202#comment-13893202
] 

Karl Wright commented on CONNECTORS-883:
----------------------------------------

Currently, the JobNotificationThread looks for jobs in the READYFORNOTIFY state.  The jobs
it finds are put in the the transient state NOTIFYINGOFCOMPLETION.  Jobs who have been notified
get put into INACTIVE state.

What this ticket proposes is the following:
- The JobManager method getJobsReadyForInactivity() should also pick up the new failTime and
failCount fields (added to the Jobs table)
- When a failure occurs, either a retry occurs (which involves putting the job back into READYFORNOTIFY
and updating failTime and failCount appropriately, the failure is skipped (the job transitions
to INACTIVE without success), or the job is marked as "failure" with the appropriate error
message and also transitions to INACTIVE.
- The only other thing that has to happen is that jobs which are placed in the INACTIVE state
MUST have their failTime and failCount fields reset appropriately.
- The failTime and failCount fields might also profitably be used during other single-thread
states which deal with ServiceInterruptions, such as seeding.



> Framework: Job notification thread does not handle notification service interruptions
properly
> ----------------------------------------------------------------------------------------------
>
>                 Key: CONNECTORS-883
>                 URL: https://issues.apache.org/jira/browse/CONNECTORS-883
>             Project: ManifoldCF
>          Issue Type: Bug
>          Components: Framework crawler agent
>    Affects Versions: ManifoldCF 1.5
>            Reporter: Karl Wright
>            Assignee: Karl Wright
>             Fix For: ManifoldCF 1.6
>
>
> When a job notification fails because of a ServiceInterruption, the notification is retried
indefinitely.  This is incorrect; the notification retry should at the very least obey the
ServiceInterruption specification pertaining to number of retries, retry interval, and failure
mode.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)

Mime
View raw message