Why is there always an extra minute of wait time after all DMAs have received the upgrade package?
This minute is significant when deploying small packages that take less than a minute to actually deploy.
Hi Matthijs,
This delay occurs on the orchestrating agent (agent through which the upgrade was initiated) after the local upload has completed, while waiting for the uploads to other agents to complete.
Once all agents have received the complete package, the prerequisite checks as present in the upgrade package will be executed, and the upgrade/upload will be marked as failed as soon as one of the agents failed on one of the prerequisites.
The line "Local upload completed. Waiting for x other agent(s) to finish uploading. (retry x/15)" is an indicator here. Currently these retries (checking if all agents completed their upload) are done at one-minute intervals.
I agree that a number of improvements are possible here to reduce or eliminate the delay. A task has been created for this on the backlog (DCP207696).
FYI: an improvement for this is already part of the upcoming changes for supporting upgrades over gRPC.
Hi Wouter,
Thanks for the answer, but are these periodic checks really needed? As the initiating agent clearly already knows that the upload was completed on 3 agents just above the red line in my screenshot?