Leader stays on EFS for now:
Every [15 minutes, if last commit has changed on the leader](look into config options), run replication?command=backup with repository configured to s3
Follower updates:
Update follower config to use local filesystem rather than EFS mount point
Update entrypoint to run replication?command=restore with repository configured to s3 (turn off standard leader/follower replication during restore)
Verify that any in-flight updates on the leader still get propagated to follower via standard leader/follower replication
Cluster updates:
Drop down to one follower ECS Service that scales to multiple containers
Update leader/follower language (vs. writer/reader language) in the code
Next Steps:
Look into the leader using local filesystem instead of EFS - restoring from the same backup in s3 - and see if we can capture and re-run any in-flight updates via the dashboard.
Leader stays on EFS for now:
Every [15 minutes, if last commit has changed on the leader](look into config options), run
replication?command=backupwith repository configured to s3Follower updates:
Update follower config to use local filesystem rather than EFS mount point
Update entrypoint to run
replication?command=restorewith repository configured to s3 (turn off standard leader/follower replication during restore)Verify that any in-flight updates on the leader still get propagated to follower via standard leader/follower replication
Cluster updates:
Drop down to one follower ECS Service that scales to multiple containers
Update leader/follower language (vs. writer/reader language) in the code
Next Steps:
Look into the leader using local filesystem instead of EFS - restoring from the same backup in s3 - and see if we can capture and re-run any in-flight updates via the dashboard.