Description
It would be nice if Nomad could implement an allocation lifecycle feature that automatically refreshes allocations after a specified number of days. Instead of having allocations run indefinitely, Nomad could replace an old allocation with a new one, keeping services fresh and avoiding issues associated with long-running allocations.
Use Cases
I have faced several issues from long-running allocations (specially on docker) such as memory leaks, stale connections, disks full of stale logs, or other surprising obstacles.
Attempted Solutions
-
Manual Restarts:
Currently, I have to manually restart allocations or trigger redeployments. This method is error-prone and can lead to downtime if not managed carefully. -
External Scheduling Tools:
Creating cron jobs or custom scripts and run it under Nomad batch job to force allocation refreshes, but these solutions are clunky and lack tight integration with Nomad’s native scheduling and resource management. This approach is also less efficient for large allocations.
Metadata
Metadata
Assignees
Type
Projects
Status