[Building Sakai] Preventing ScheduledInvocationRunner from running on a node

[Zhen Qian]

Hi, Chris:

Given that we have had recent hiccups with job scheduler, this feature
would be helpful to us, too.

Thanks,

Thanks,




On Mon, Mar 25, 2013 at 12:44 PM, Maurer, Christopher Wayne <
chmaurer at iupui.edu> wrote:

> We have been running something at IU for several years that lets us
> configure which server nodes (in a cluster) will register themselves to
> run jobs. I imagine that we'd be happy to contribute it back.
>
> Chris
>
> On 3/22/13 3:00 AM, "Steve Swinsburg" <steve.swinsburg at gmail.com> wrote:
>
> >Yep sounds reasonable. We need to be flexible for clusters.
> >Cheers
> >Steve
> >
> >Sent from my iPhone
> >
> >On 22/03/2013, at 3:23, David Adams <da1 at vt.edu> wrote:
> >
> >> We ran into similar problems when we tried to add an "admin" server to
> >> our cluster with a different URL. The practice of background jobs
> >> self-registering and being completely unmanageable is very problematic.
> >> Seems like at the very simplest, you could just add a property that
> >> would allow disabling this job on a per-server basis. Sample patch
> >> attached.
> >>
> >> I can't see any reason why disabling this on one or a few nodes would
> >> be a problem so long as it's running somewhere.
> >>
> >> -dave
> >>
> >> Seth Theriault wrote:
> >>> Hello,
> >>>
> >>> By default, a Sakai server registers a ScheduledInvocationRunner job
> >>> at startup. This job runs every 10 minutes and queries the
> >>> SCHEDULER_DELAYED_INVOCATION table for tasks to be executed. Reminders
> >>> and notifications from Evaluation, Forums, and Content are done this
> >>> way.
> >>>
> >>> I am wondering if there is any danger in NOT registering this job on a
> >>> single node in a cluster of Sakai servers. Locally, we have a "feeds"
> >>> server that is not user-facing but accesses the same database and
> >>> other data -- including this table -- as the other servers. People are
> >>> getting reminders from this server -- which has a different URL -- and
> >>> it's confusing.
> >>>
> >>> Looking at:
> >>>
> >>>
> >>>jobscheduler/scheduler-component-shared/src/java/org/sakaiproject/compon
> >>>ent/app/scheduler/ScheduledInvocationManagerImpl.java
> >>>
> >>> there doesn't currently seem to be a method to prevent
> >>> auto-registration of the job on a particular node, but I think that
> >>> could be easily done.
> >>>
> >>> Opinions?
> >>>
> >>> Seth
> >> <disable-sir.patch>
> >> _______________________________________________
> >> sakai-dev mailing list
> >> sakai-dev at collab.sakaiproject.org
> >> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
> >>
> >> TO UNSUBSCRIBE: send email to
> >>sakai-dev-unsubscribe at collab.sakaiproject.org with a subject of
> >>"unsubscribe"
> >_______________________________________________
> >sakai-dev mailing list
> >sakai-dev at collab.sakaiproject.org
> >http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
> >
> >TO UNSUBSCRIBE: send email to
> >sakai-dev-unsubscribe at collab.sakaiproject.org with a subject of
> >"unsubscribe"
>
> _______________________________________________
> sakai-dev mailing list
> sakai-dev at collab.sakaiproject.org
> http://collab.sakaiproject.org/mailman/listinfo/sakai-dev
>
> TO UNSUBSCRIBE: send email to
> sakai-dev-unsubscribe at collab.sakaiproject.org with a subject of
> "unsubscribe"
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://collab.sakaiproject.org/pipermail/sakai-dev/attachments/20130325/1eec8c29/attachment.html