Remove outdated TODO from orchestration module
This commit is contained in:
parent
3c2c2f2255
commit
7d5561d639
@ -1,29 +0,0 @@
|
||||
1. Core orchestration
|
||||
1.1. Celery integration
|
||||
1.2. Orchestrate stuff based on plan traversal
|
||||
1.3. Different controls
|
||||
1.4. Granular execution (e.g execute start/end, execute certain tasks, execute certain path)
|
||||
|
||||
2. User facing interface for orchestration
|
||||
2.1. How to integrate this stuff with resources?
|
||||
Two options:
|
||||
- Use orchestrator as driver
|
||||
- Orchestrate resources externally, all logic implemented in resource driver
|
||||
2.2. How to allow reuse across different Profiles ?
|
||||
Solution: plan for each separate group of resources, e.g. plan for rabbitmq
|
||||
deployment is not exposed, it is stored only in rabbitmq resources template,
|
||||
but it exposes several *anchors* - amqp_cluster_ready and amqp_one_node_ready
|
||||
|
||||
3. Granular testing
|
||||
3.1. How to integrate pre/post verifications with graph execution
|
||||
|
||||
4. Add back timeout support
|
||||
|
||||
Orchestration features
|
||||
-------------------------
|
||||
1. Controls
|
||||
- Execute task even if all predecessors failed
|
||||
- Execute task when one (specific number) predecessor succeed
|
||||
- Execute task when certain percent of tasks are success
|
||||
- Ignore if some task failed
|
||||
2. Granular execution
|
@ -11,6 +11,7 @@ ERROR - visited node, but failed, can be failed by timeout
|
||||
SUCCESS - visited node, successfull
|
||||
INPROGRESS - task already scheduled, can be moved to ERROR or SUCCESS
|
||||
SKIPPED - not visited, and should be skipped from execution
|
||||
NOOP - task wont be executed, but should be treated as visited
|
||||
"""
|
||||
|
||||
|
||||
|
Loading…
x
Reference in New Issue
Block a user