Hello! We want to have possibility to continue the...
# ask-communityv
Vadym Dytyniak
06/22/2023, 1:45 PMHello! We want to have possibility to continue the flow run in case K8S pod was killed and flows run restarted on the new pod. Is it possible to use some Prefect features to share that state and results from executed tasks to continue the flow run from where it was interrupted?
✅ 1
c
Christopher Boyd
06/22/2023, 3:04 PMYou can persist results to remote storage, or an underlying storage mount
Christopher Boyd
06/22/2023, 3:05 PMIt would be the same concept as if you did this locally in your environment, with the caveat that pods are ephemeral, so you would need to configure the result persistence manually
v
Vadym Dytyniak
06/23/2023, 9:49 AMCould you please explain how it works in more details, because I can't find details in documentation? I see a lot of arguments that I can pass in
@taskc
Christopher Boyd
06/23/2023, 4:58 PMChristopher Boyd
06/23/2023, 4:58 PMChristopher Boyd
06/23/2023, 4:58 PM Copy code
If there is no context to load the storage from and results must be persisted, results will be stored in the path specified by the PREFECT_LOCAL_STORAGE_PATH setting (defaults to ~/.prefect/storage).Christopher Boyd
06/23/2023, 4:59 PMsince pods are transient,
/.prefect/storageChristopher Boyd
06/23/2023, 4:59 PMso you need to persist your storage either remotely, or using a volumeMount in kubernetes for provisioned storage
🙌 1
Christopher Boyd
06/23/2023, 5:00 PMv
Vadym Dytyniak
06/26/2023, 12:14 PMThanks. Could you please explain how second attempt will know how to find needed results? Will it use flow run uuid + task id?