The point of those directories isn't input, its to dictate when a job runs. The directories only create a dependency between jobs.
-- Chris On Tue, Apr 23, 2013 at 7:32 AM, Tousif <[email protected]> wrote: > My job is a crawler and it doesn't depend on input . > > > On Tue, Apr 23, 2013 at 4:51 PM, Chris Sigman <[email protected]> wrote: > > > One way you might be able to do this is have the coordinator have an > input > > event path instance be timestamped to the minute and have the coordinator > > job run every minute as well. Then, the last thing the job will do before > > finishing is to create one of those timestamped paths with the current > > minute + a safety offset (maybe just a minute). As an example, job X > > finishes at 1:23 AM and creates a directory named 0124. Job Y was > > scheduled to run at 1:24 AM. When it starts, it sees the directory > > present, so it executes. Job Z was scheduled to run at 1:23 AM, but a > > directory named 0123 was never created, so it never executes. > > > > > > -- > > Chris > > > > > > On Tue, Apr 23, 2013 at 5:22 AM, Tousif <[email protected]> wrote: > > > > > Hi, > > > > > > I would like to run a coordinator job in a loop. As soon as existing > job > > > finishes it > > > should start same job again , Here job runs for unpredictable time. > > > > > > -- > > > > > > > > > Regards > > > Khazi > > > > > > > > > -- > > > Regards > Tousif Khazi > > . > "A player that makes a team great is more valuable than a great player". >
