-
Feature Request
-
Resolution: Won't Do
-
Normal
-
None
-
2.4, 2.5
-
False
-
-
False
- What is the nature and description of the request?
When an execution node wants to use an EE whose pull policy is "Always pull", and the following conditions are met:
- the `podman pull` command fails for any reason, and
- the execution node already has some version of the EE image stored locally,
THEN: a failure to `podman pull` should be worked around and the container should run with the existing version.
A clear WARNING should still be issued in case of `podman pull` failures so this doesn't become a silent failure. But `podman pull` failures should not prevent jobs from running if possible.
- Why does the customer need this? (List the business requirements here)
The `podman pull` command may fail for a number of reasons, from networking issues to Private Automation Hub misconfiguration. EE images with a policy of "always pull" will often detect that there's no change in the image so nothing will be really pulled. Occasional failures to `podman pull` should definitely be investigated by AAP and PAH admins, but they should not prevent an otherwise perfectly healthy AAP from running jobs. - How would you like to achieve this? (List the functional requirements here)
Create a new download policy named "try to pull" that is a variant of "always pull": it will always attempt to `podman pull` the image and will use an existing local version of the EE image if `podman pull` fails and a local image exists. - List any affected known dependencies: Doc, UI etc..
Doc, UI, API.
- links to
