987d6a346f
When using amphorav2+persistence, taskflow maintains the ownership of a task using a TTL in redis for each job. When a task is longer than expected, it is released in redis and another controller can take it over while it is still running on the first controller. Add a keepalive mechanism in the worker to periodically reset the TTL while the worker is still processing a job Story 2009761 Task 44242 Change-Id: I6619ba117f7051fe44086389789bc6e2810fd23d (cherry picked from commit f13c3ca6b8480792e2ab86a5413843b56eb453c6)
9 lines
358 B
YAML
9 lines
358 B
YAML
---
|
|
fixes:
|
|
- |
|
|
Fix an issue with amphorav2 and persistence, some long tasks executed by a
|
|
controller might have been released in taskflow and rescheduled on another
|
|
controller. Octavia now ensures that a task is never released early by
|
|
using a keepalive mechanism to notify taskflow (and its redis backend) that
|
|
a job is still running.
|