On 11/26/2014 02:50 PM, Yaniv Dary wrote:
----- Original Message -----
From: "Yair Zaslavsky" <[email protected]>
To: "Sahina Bose" <[email protected]>
Cc: [email protected]
Sent: Wednesday, November 26, 2014 7:32:32 AM
Subject: Re: [ovirt-devel] Certain questions about Quartz based scheduling in
oVirt
----- Original Message -----
From: "Sahina Bose" <[email protected]>
To: "Moti Asayag" <[email protected]>, "Shubhendu Tripathi"
<[email protected]>
Cc: [email protected]
Sent: Tuesday, November 25, 2014 1:13:06 PM
Subject: Re: [ovirt-devel] Certain questions about Quartz based scheduling
in oVirt
On 11/24/2014 06:27 PM, Moti Asayag wrote:
Hi Shubhendu,
----- Original Message -----
From: "Shubhendu Tripathi" <[email protected]>
To: [email protected]
Sent: Monday, November 24, 2014 8:58:34 AM
Subject: [ovirt-devel] Certain questions about Quartz based scheduling
in
oVirt
Hi All,
We are in a requirement where we need to schedule jobs at certain time
interval, hourly, daily, weekly and monthly (i.e. repetitive and cron
kind of scheduling).
I was trying to understand quartz based scheduling mechanism in oVirt to
achieve the scenarios.
Have some basic questions regarding the same -
1. Is there is mechanism to persist the scheduling data in oVirt ?
In ovirt we do not persist the jobs. The application reschedule the jobs
when it starts
and programmatically triggers jobs when required.
On packaging/services/ovirt-engine/ovirt-engine.xml.in we specify the job
store configuration
as RAMJobStore, which is a volatile:
<property name="org.quartz.jobStore.class"
value="org.quartz.simpl.RAMJobStore"/>
You may select other implementation. See:
http://quartz-scheduler.org/api/2.2.0/org/quartz/spi/JobStore.html
Would the correct approach then be to have multiple scheduler instances?
Please bare in mind that quartz has the ability to perform
Cron like scheduling (with cron expressions).
http://www.quartz-scheduler.org/documentation/quartz-1.x/tutorials/crontrigger
+1
I'm don't think creating a cron like ability in RHEV is needed this is why
cron/quartz exists.
Instances for users jobs should be added independently from RHEV.
Yes, we would use the quartz cron scheduling capabilities.
The point under discussion is the job persistence - to use Quartz's DB
persistence feature, so that we do not have to implement the logic of
instantiating these jobs on server restart
And if we need multiple scheduler instances - one that uses the volatile
RAMJobStore, and another that uses the persistent database job store
(JobStoreCMT)
thanks!
sahina
1. - that uses the in memory job store
2. that uses DB to persist the jobs.
The second instance would be used to schedule and manage any dynamic
jobs, for instance like the ones required for gluster volume snapshot
scheduling
Using a separate instance also would mean there's no change to existing
jobs that are scheduled in Backend bean.
2. How to tackle edit and rescheduling of jobs ?
You can have a look at org.ovirt.engine.core.utils.timer.SchedulerUtil
interface in
the 'scheduler' project which provides that functionality, and the
related
classes.
But if not special requirements, i guess the shipped quartz
implementation
should be
enough:
http://quartz-scheduler.org/documentation/quartz-2.x/tutorials/tutorial-lesson-05
Kindly guide on how to achieve these.
Thanks and Regards,
Shubhendu
_______________________________________________
Devel mailing list
[email protected]
http://lists.ovirt.org/mailman/listinfo/devel
_______________________________________________
Devel mailing list
[email protected]
http://lists.ovirt.org/mailman/listinfo/devel
_______________________________________________
Devel mailing list
[email protected]
http://lists.ovirt.org/mailman/listinfo/devel
_______________________________________________
Devel mailing list
[email protected]
http://lists.ovirt.org/mailman/listinfo/devel
_______________________________________________
Devel mailing list
[email protected]
http://lists.ovirt.org/mailman/listinfo/devel
