java - JSR 352 Restarting a job? -
when job restarted, failed partitions run again. how run completed job again?
is there difference between submitting job again , restarting job?
i using ibm's implementation of jsr 352 on websphere liberty.
java batch designed when restart job instance, execution continues left off (in previous failed or stopped execution).
so typically means 2 things:
- within job, begin @ previously-failing step (or step @ job stopped).
- within step, begin positioning input cursor using checkpoint values kept batch container.
so if on initial execution completed step1 failed @ step2 @ record #4123, , checkpointing every 100 records, on restart you'd typically begin executing @ step2 @ record #4100.
in cases it's necessary execute step1 on restart well, before executing step2 job had failed, , there option behavior well.
it typical submit/start job repeatedly, on schedule.
in terms of batch specification, new jobinstance created each time job started. @ time initial jobexecution jobinstance created. if execution not run completion, instance may restarted, @ time second jobexecution created jobinstance.
so restart uses checkpoints , execution history of previous jobexecution while start begins clean slate.
Comments
Post a Comment