Compute grid jobs will execute more than once if the server crashed in the middle of execution. In that case, the job will be failed over and started from scratch.
If you want exactly once guarantee, then assign IDs to your computations or events and store them in a cache. This way, if the event ID is already there, then the job has already executed. You can do such cache updates within a transaction in order to make sure that everything happens atomically. Does this make sense? D. On Tue, May 17, 2016 at 6:19 PM, ght230 <[email protected]> wrote: > According to the reference document, ignite compute grid can provide at > least > once guarantee. > > I want to know in what conditions will compute grid execute the jobs more > than once and will it lead to incorrect results. > > > > -- > View this message in context: > http://apache-ignite-users.70518.x6.nabble.com/Can-compute-grid-guarantee-the-correctness-of-the-result-tp4992p5006.html > Sent from the Apache Ignite Users mailing list archive at Nabble.com. >
