Thanks.
1. The "non-cancellation" behavior is useful.
2. Our worker results tend to be heavyweight  we would ideally like another
master to transparently take over the role of the failed master without the
worker knowing about master node failure and without having to save
results. Is this possible?

On Mon, Aug 29, 2016 at 4:27 PM, vkulichenko <[email protected]>
wrote:

> Vinay,
>
> If job implements ComputeJobMasterLeaveAware it will not be cancelled when
> the master node fails. It also gives you a callback that notifies you about
> the event so that you can save the results to be collected by something
> else. You can save results in cache or use checkpoints [1].
>
> [1] https://apacheignite.readme.io/docs/checkpointing
>
> -Val
>
>
>
> --
> View this message in context: http://apache-ignite-users.
> 70518.x6.nabble.com/How-to-accomplish-Master-node-
> failover-without-resubmit-tp7373p7386.html
> Sent from the Apache Ignite Users mailing list archive at Nabble.com.
>

Reply via email to