2021-05-20 02:51:22 UTC - Amndeep Singh Mann: Hi folks,
Would appreciate help with figuring out how to deal with an unsuccessful
deployment and subsequent unsuccessful undeployment of some functions.
Situation:
1. Deployed functions successfully from a manifest file.
2. Reconfigured and redeployed the openwhisk server to reduce the maximum
timeout value from what we had originally increased it to.
3. Forgot to reduce the maximum timeout value in the manifest file accordingly
and attempted to deploy again.
4. Deploy failed due to an error stating that the manifest was malformed (kinda
misleading term imo considering this exact file worked a few minutes
previously) due to the timeout value being too large.
5. Reduced the timeout value in the manifest file and attempted to deploy
again, which failed due to the partially deployed assets existing from the
previous attempted deployment.
6. Tried a whole bunch of stuff to undeploy the functions which failed due to
“Resource by this name exists but is not in this collection.” and “Package not
empty” errors.
a. Attempted to use the undeploy subcommand
b. Attempted to change the manifest file to the “malformed”/original
variant and undeploy
c. Attempted to delete the package manually
d. Attempted to delete the functions manually
e. Wanted to try to use the ‘force’ parameter for the undeploy/deletes, but
the wsk cli does not support that option as far as I could tell/attempted
7. Gave up for the moment, changed the package name in the manifest to
something else and successfully deployed the functions again.
Is there a way to clean up these ‘zombie’ deployed functions?
https://openwhisk-team.slack.com/archives/C3TPCAQG1/p1621479082010800?thread_ts=1621479082.010800&cid=C3TPCAQG1
----
