Print

[nabilasbi]

Hello,

I'm trying to understand how JPPF act when one of its components crash, and how recovery of the driver and nodes work in case it was running a job before it crashed , Please advise since I cant find in the documentation what I need?

Thank you

[lolo]

Hi Nabil,

The behavior in case of a server crash is explained in details in this part of the documentation.

When a node crashes, the tasks that it was executing are put back in the server's queue, so they can be disptached to another (or more) node.

In the case of a client crash, there are 2 possibilities for the jobs submitted:
- you can tell the server to continue executing the jobs, and the results will be lost. This is configured on each job, via JPPFJob.getSLA().setCancelUponClientDisconnect(false);
- you can use a persistence manager, so that any results received while the client was alive will be persisted and restored when the client is restarted. We have a sample showing how this can be used here.

-Laurent