Currently in ecflow we can have jobs that have identical but vary only in the step.
In these cases each step carries a latency, i.e submission cost,(i.e. starting hundreds of new processes)
The Queue attribute was created to address this situation.
Code Block |
---|
suite test_queue
family f1
queue q1 1 2 3 4 5 6 7
task t
endfamily
family f2
task a
queue q2 1 2 3
task b
trigger /test_queue/f1:q1 > 1
task c
trigger /test_queue/f2/a:q2 > 1
endfamily
endsuite |
There are new child commands, that are used to signal when a step is active,complete or has aborted.
Code Block | ||||
---|---|---|---|---|
| ||||
step=$(ecflow_client --queue queue_name active) # This will return next step form the server and make it active
ecflow_client --queue queue_name complete # complete the current active step
ecflow_client --queue queue_name aborted # tell server that current step failed. |