You are viewing a plain text version of this content. The canonical link for it is here.
Posted to oak-issues@jackrabbit.apache.org by "Marcel Reutegger (JIRA)" <ji...@apache.org> on 2017/08/31 09:46:00 UTC
[jira] [Updated] (OAK-6605) Provide job name for async index update
[ https://issues.apache.org/jira/browse/OAK-6605?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Marcel Reutegger updated OAK-6605:
----------------------------------
Description:
Starting with Sling Scheduler 2.6.0 runOn=SINGLE jobs are not scheduled exclusively on the leader anymore (SLING-5387). The implementation will try to distribute them evenly on all cluster nodes. The placement of these jobs depends on the name of the job, which is currently not set by Oak for the async index updates. In this case Sling will generate a name using the service id, but this is not guaranteed to be the same across all cluster nodes.
This means, it may happen that an async index update job is not scheduled at all, because all cluster nodes assume some other cluster node will schedule the job. Oak should therefore provide a name for the async index update jobs it schedules to help Sling identify those jobs.
Please note, there is a problem with the initial implementation of this feature in Sling, which got fixed in 2.7.0: SLING-7037
was:
Starting with Sling Scheduler 2.6.0 runOn=SINGLE jobs are not scheduled exclusively on the leader anymore. The implementation will try to distribute them evenly on all cluster nodes. The placement of these jobs depends on the name of the job, which is currently not set by Oak for the async index updates. In this case Sling will generate a name using the service id, but this is not guaranteed to be the same across all cluster nodes.
This means, it may happen that an async index update job is not scheduled at all, because all cluster nodes assume some other cluster node will schedule the job. Oak should therefore provide a name for the async index update jobs it schedules to help Sling identify those jobs.
Please note, there is a problem with the initial implementation of this feature in Sling, which got fixed in 2.7.0: SLING-7037
> Provide job name for async index update
> ---------------------------------------
>
> Key: OAK-6605
> URL: https://issues.apache.org/jira/browse/OAK-6605
> Project: Jackrabbit Oak
> Issue Type: Improvement
> Components: core
> Reporter: Marcel Reutegger
> Assignee: Marcel Reutegger
> Priority: Minor
> Fix For: 1.7.7
>
>
> Starting with Sling Scheduler 2.6.0 runOn=SINGLE jobs are not scheduled exclusively on the leader anymore (SLING-5387). The implementation will try to distribute them evenly on all cluster nodes. The placement of these jobs depends on the name of the job, which is currently not set by Oak for the async index updates. In this case Sling will generate a name using the service id, but this is not guaranteed to be the same across all cluster nodes.
> This means, it may happen that an async index update job is not scheduled at all, because all cluster nodes assume some other cluster node will schedule the job. Oak should therefore provide a name for the async index update jobs it schedules to help Sling identify those jobs.
> Please note, there is a problem with the initial implementation of this feature in Sling, which got fixed in 2.7.0: SLING-7037
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)