[Building Sakai] Preventing ScheduledInvocationRunner from running on a node

[Thomas, Gregory J]

I'll try to get IU's version of this before the end of the day.

They're really two separate changes, but I'll combine it into one for the
patch.

1. In sakai.properties (or whatever properties file you want to use)
you'll configure how many servers and specify which ones you want to allow
to run quartz jobs on your cluster, e.g.

## property to enable quartz only on a select number of servers ##
quartz.serverId.count=2
quartz.serverId.1=nameofoneofyourservers
quartz.serverId.2=nameofoneofyour2ndserver


2. Building off of this functionality, the 2nd fix only allows servers
defined in the above property to register the ScheduledInvocationRunner
job on startup, and I think there was some logic added to prevent
duplication.  One of the issues at IU we had in production is our 12
server cluster would repeatedly unregister and sometimes register
ScheduledInvocationRunner.  Since the servers start up in roughly the same
timeframe, this caused issues with this process.  If the job isn't there,
email notifications are not sent out when they're suppose be sent out.  IU
currently allows just one server on our cluster to be the "quartz server".

We've had both of these in our production environment for at least 1 year
(maybe 2-3).

I'll send an email when the patch is ready.

Thanks,
Greg



On 3/25/13 1:32 PM, "Seth Theriault" <slt at columbia.edu> wrote:

>On Mon, Mar 25, 2013 at 1:05 PM, Maurer, Christopher Wayne
><chmaurer at iupui.edu> wrote:
>> I'll see if we can get a patch together sooner rather than later.
>
>I have opened a Jira for this:
>
>https://jira.sakaiproject.org/browse/SAK-23415
>
>David Adams's original patch is attached there for reference.
>
>Seth
>_______________________________________________
>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"