Hey all - Is there a way to make archived version ...
# ask-community
k
Hey all - Is there a way to make archived version of the flow active again? I have a flow that runs on a K8s agent, the image is stored in a container registry. I push new version to production and find a bug - I delete this new version. How do I now run the previous version of the flow, w/o creating a new Docker image?
🙌 1
h
Actually came here today to ask the same question!
Can I revert to a previous version of the flow?
k
Great - I am sure we are not the only ones asking for this feature 🙂
k
Hey @Klemen Strojan, at this time you can’t because we don’t have guarantees that the version of that code will still exist (think Github file changes or S3 file changes). But we are working on two things that may be of interest. First is that you will be able to supply your own version rather than have it just be a number that keeps going up. You’ll be able to supply a string. The second thing is that you will be able to have multiple versions of a Flow at a given time.
k
Thanks @Kevin Kho, that makes sense - thanks for your response 🙂
e
@Kevin Kho so until the second part is done, what can I do with old versions of the flow other than see the DAG?
k
Honestly nothing other than check the previous flow runs and filter by version in the UI or API. This is why we have rethought it moving forward. This was kinda covered in the Orion release. Have you seen that?
e
k
Yep!
e
I read it very briefly and havn’t seen something related to the versioning, i’ll read it in depth
k
Ah the gist is that you provide versioning and it can be anything (like a hash or a number of string) and you will be able to have multiple versions of the same flow at a given time that you can run in Orion
e
amazing! so if I have a new version of the flow, the old flow are still reachable and hopefully their schedulers will still run on that flow and not the new one
k
I believe this is right you can have concurrent versions of the same flow. This is intended to support cases where users have sets of parameters that they want to consistently run, but the flow logic is the same
🙌 1