"SPIKE: Test running backup as part of the pre-deployment step"

As part of refining #42 (closed) we discussed also adding a gcloud command to snapshot the db prior to any migrations being attempted to protect against failure.

This issue is to implement this on-demand backup requirment using the gcloud sql backups create command.

added teamCloud workflowSprint Ready labels

changed iteration to Cloud Team Sprints Jun 7, 2023 - Jun 20, 2023

assigned to @rk725

marked this issue as blocked by #42 (closed)

added workflowBlocked label and removed workflowSprint Ready label

removed iteration Cloud Team Sprints Jun 7, 2023 - Jun 20, 2023

changed iteration to Cloud Team Sprints Jun 21, 2023 - Jul 4, 2023

added workflowSprint Ready label and removed workflowBlocked label

added workflowNeeds Refinement label and removed workflowSprint Ready label

Sending back to refinement as the solution to the migration issue ended up being vastly different to the original plan. Need to (re-)discuss how we'd like to implement this issue.

removed iteration Cloud Team Sprints Jun 21, 2023 - Jul 4, 2023

From Refinement:

• There will need to be an option to skip the db backup. It would be good if the code to autodetect if the database existed.
• Make sure this doesn't increase the job time too much when running.

added workflowSprint Ready label and removed workflowNeeds Refinement label

added priority2 Medium label

changed iteration to Cloud Team Sprints Nov 22, 2023 - Dec 5, 2023

Discussion from Refinement:

• Should there be a different pipeline called backup-and-migrate?
• Can we trigger a db snapshot which is triggered when the CloudRun image is deployed

added workflowNeeds Refinement label and removed workflowSprint Ready label

removed iteration Cloud Team Sprints Nov 22, 2023 - Dec 5, 2023

changed title from terraform-pipeline.yml: add db snapshot as part of the apply job to protect against issues with migrations to Add db snapshot as part of the apply job to protect against issues with migrations

Refinement:

• This should be turned in to a spike to test if this works as part of the pre-deployment step
• If so, then we should document and communicate to teams
• How long do backup stay?
• If there a performance/cost implication about having a db snapshot on every deploy?