Hi,
I have a single nomad client which is a macOS machine. I submitted a “batch” job that downloads and runs a binary with some arguments for me.
Once the batch job finishes, it goes into the completed state. Now, my nomad client i.e. the mac kinda goes into screen lock – making the state of the client as “disconnected” as shown below. This is because my job had the following parameter passed: max_client_disconnect = "1h"
What happens next is that the job is kinda scheduled again – which I don’t want since it was already completed before. We can see that it goes into pending as seen below.
Once I unlock my mac again i.e. the client becomes Ready, the job which was pending successfully runs again. This basically means the job is run again even though it has completed successfully before.
One can confirm this by checking the 2 allocations listed below – one is 25 mins ago while the other is just a few seconds ago – both of which are completed.
How can I stop this behavior? I am already using the following block for my group:
max_client_disconnect = "1h"
prevent_reschedule_on_lost = true
reschedule {
attempts = 0
unlimited = false
}
restart {
attempts = 0
mode = "fail"
}
but it seems to have no effect whatsoever 
TLDR: a batch job which is completed is retried again if the nomad client goes down (say screen lock) for a while and then comes back up — I don’t want this.