ru-fix / distributed-job-manager

DistributedJobManager will balance workload between servers for you.
https://ru-fix.github.io/distributed-job-manager
3 stars 1 forks source link

Allow worker to continue his jobs executions during connection suspended #65

Open KazankovMarch opened 4 years ago

KazankovMarch commented 4 years ago

Sometimes connection is suspended for a moment and then reconnected. Despite of tiny duration of connection suspention, worker launches shutdown of all jobs and manager launches rebalance. Maybe we should allow worker to continue its work during some timeout in hope that it will reconnected to zk. So manager should consider that worker isn't dead during this timeout.