Quartz.net Clustering Guide

[webdawg]

Hi Everyone,

I have Quartz working now and want to move onto configuring the
clustered instanced on our other servers.

I've looked, but cannot find much info. Is there a guide anywhere?

As far as I can tell there is only one variable to set to make them
clustered and then you just need to have the server times in sync.

However, I still have questions, such as:

1. Is one instance the parent and another takes over if the parent
goes down? Or do they all sort of fire off at the same time, but
somehow manage not to have duplicates?

2. Is there a recommended procedure for updating our class
libraries? e.g. should all instances be turned off first... should
we turn them back on in a specific order?


Obviously, I can install it on several servers and play around, but
that is just guess work and we may miss critical steps. I'm hoping
someone that did this before can help point me in the correct
direction.. We really appreciate the help!

[Jay Vilalta]

For me it's easiest to think of clustering as if they are all basically
independent instances that share one job store. Any instance can run any job
that is fired at any time, but Quartz.Net doesn't let the same job fire
twice on different servers.

Regarding updates, it depends what changes you made to your libraries.
Safest would be to turn them all off, update them and start them up, but if
you can't do that you'll have to think through what might happen if, for
example, you created a new type of job and a server that hasn't been updated
tries to execute a job type that it doesn't have yet. You'll have to
consider exactly what changed to determine if it's safe to do incremental
updates.

--
You received this message because you are subscribed to the Google Groups
"Quartz.NET" group.
To post to this group, send email to quar...@googlegroups.com.
To unsubscribe from this group, send email to
quartznet+...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/quartznet?hl=en.

[Matt Weber]

Hi Jay, 

Thank you!  That is exactly what I needed to know.

[Alexander Popov]

Can i  for this create Scheduler,  but not call scheduler.start() ?

just use scheduler.scheduleJob()

On Thursday, October 11, 2012 8:40:38 PM UTC+4, Alexander Popov wrote:

How can i make one( or several) node to only take jobs and put in storage but not do any work.
Setup that i want to build:

1. several Nodes with simple windows service that host

2. one or several servers with just REST API to take jobs and putting  it into storage, reporting about job status.

3. clients goes to REST servers  and put|check job status.

Now i not understand how to setup 2.  Just create scheduler when request is coming put job and shutdown?

[Drew Mace]

Alex,

Do not call .Start().

Also in the node that will only take jobs, use the following:

properties["quartz.threadPool.type"] = "Quartz.Simpl.ZeroSizeThreadPool, Quartz";

Hope this helps.