is there any way to fallback the swithover of RDS blue/green deployment?

0

To process the EOL of aurora MYSQL 5.6, i need upgrade the major version.

here is my case, i have a database named avocado(the blue one), and i create a blue/green deployement and swith the traffic to green one(avocado-xxx), but right now i find some issues in green cluster, so i want to switch traffic back to the blue cluster, i know the data-loss from switching time. is there any way to let me switch back to blue qucikly? or can i restore in green cluster using the snapshot which i create before switching in blue cluster?

Looking forwarding to get your information, Thanks team and best regards.

asked a year ago2602 views
3 Answers
1

Hi

Please, take a note that the green cluster stays in sync with the blue one using logical replication. [1]

This is the reason why binary logging is required for replication from the blue environment to the green environment. [2]

The switchover process works by promoting a DB cluster in the green environment. [3]

Replica promotion works by stopping replication process between a source and a target and promoting a target to independent DB instance.

Therefore, after the switchover there are 2 completely independent DB instances.

The quickest way to switch back to the former blue cluster in your case would be to rename the former blue/green DB cluster identifiers:

  • Modify the "DB cluster identifier" of the former green instance to some other name; [4]
  • Modify your former blue "DB Cluster Identifier" to the name it had before. [4]

These steps would change the former blue cluster endpoints to the ones used before.

Make sure to apply the changes immediately.

I hope that the information provided has been helpful, and I was able to address your query.

References:

[1] Overview of Amazon RDS Blue/Green Deployments for Aurora - https://docs.amazonaws.cn/en_us/AmazonRDS/latest/AuroraUserGuide/blue-green-deployments-overview.html
[2] Creating a blue/green deployment - https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/blue-green-deployments-creating.html [3] Switching a blue/green deployment - https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/blue-green-deployments-switching.html
[4] Modifying an Amazon Aurora DB cluster - https://docs.aws.amazon.com/AmazonRDS/latest/AuroraUserGuide/Aurora.Modifying.html

AWS
answered a year ago
  • This is a good insight. I wonder if, after the switchover, it would be possible to enable binary logging on OLD- BLUE to keep it in sync with NEW-BLUE. Basically, making OLD-BLUE a new GREEN environment, without data loss. This would enable quick fallback and enable disaster recovery

0

I think we can't fallback to the blue one with blue/green feature at this time.

After switching over, blue and green databases are separated databases so we have to treat it as a separate database. If we want to use the old (blue) one, we have to connect the old endpoint manually.

I also hope the Blue/Green fallback feature is coming :)

https://docs.aws.amazon.com/ja_jp/AmazonRDS/latest/UserGuide/blue-green-deployments-switching.html#blue-green-deployments-switching-over

https://aws.amazon.com/jp/blogs/aws/new-fully-managed-blue-green-deployments-in-amazon-aurora-and-amazon-rds/

fu3ak1
answered a year ago
  • Thanks firstly, ok, so i can not easily switch back to old cluster. i get the point. But i have another question, if i created a snapshot from old one before switch, can i restore the new cluster using the snapshot. my concern is the snapshot come from old one/version 5.6, and new one is 5.7.

0

It is not entirely clear what you are asking. When you create a blue green deployment, RDS will create an avocado-green cluster and keep it up to date with avocado(blue) via binlog replication. At anytime you can delete the green cluster or delete the blue green deployment. This will have no impact on blue. However once you call switchover, RDS will swap the roles from blue to green. And as part of the switchover stop the replication to green. After switchover blue will be renamed avodaco-old, and it will start to diverge from avocado (green). You can still go back and continue to use the original cluster, but there is no automation to switch back to avodaco-old.

AWS
MODERATOR
philaws
answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions