Fixes
engine: a job without a Nomad deployment, like a batch job, no longer
fails its deployment on the first failed or lost allocation when Nomad
reschedules it. The replacement decides, within the apply timeout, and the
deployment fails only once Nomad gives up on the allocation
(healthy,
#277).engine: a job that Nomad couldn't read or plan no longer looks gone from
git until the next cycle. It shows the new Nomad error sync state with
the error, an approval answers409with it, and pause and resume work.
nops_job_driftandnops_job_blockedhave no series for it
(sync state of a job,
metrics,
#278).
Commits: v0.15.1...v0.15.2