Query on upgrading gocd server 19.12 running in a docker container

[Steve Howe]

Hi,

First time posting, apologies for any etiquette oopsies. I've searched for answers for this but can't find anything obvious, google-fu is failing me today.

I have a question. I have a copy of GoCD (gocd/gocd-server:v19.12.0) server running in a docker container on an AWS EC2 instance with the GoCD directory mounted into the GoCD docker container into the expected GoCD directory from a mounted EFS share on the EC2 instance (using the embedded database currently). The idea of the EFS directory mounted into the container is that the EFS should persist if the container crashes or the EC2 instance dies, it's re-attached to the new ASG'd EC2 instance when it is started up.

Does anyone have any advice please, how I should go about upgrading to v20.5?

Possible options I can think of:

Maintenance mode, take a backup, get it off-server, then:

"docker exec" into the GoCD 19.12 docker container, download the v20.5 upgrade setup file and run it. This will upgrade the GoCD version in the docker container and also upgrade the GoCD directory mounted into the container, in the EFS share. Once complete, the v19.12 (now v20.5) docker container would be stopped, erased, a new, clean v20.5 docker container that matches the upgraded GoCD data directory would be started. Then, a Postgres RDS would be setup and the embedded DB migrated out to the Postgres DB.

OR

Stop the GoCD v19.12 docker container, erase it, start a new GoCD v20.5 docker container (with the EFS GoCD data directory mounted into it) and hope that the new docker container will automatically make the appropriate changes in the GoCD directory mounted into the container. Then, a Postgres RDS would be setup and the embedded DB migrated out to the Postgres DB.

Any ideas on how I should progress please?

Thanks in advance for any wisdom,

Steve

[Aravind SV]

Hello Steve,

I’m not the best person to answer this. I think there are others here who would be using GoCD in this way. I would lean heavily towards approach 2, and not messing around with docker exec, etc.

You should still take a full backup before you do anything. You should be able to copy everything that matters (DB, config) and try out the second approach you mentioned, without too much risk I think.

Cheers,
Aravind

[Steve Howe]

Will have a crack at that, thanks @Aravind!

Steve