External Locking DB: Redis #2491
Conversation
This comment was marked as resolved.
This comment was marked as resolved.
|
Absolutely Cool @SudoSpartanDan! What a step up for Atlantis. Thank you for the efforts! |
jamengual
added
feature
|
Another nit. There are many rearrangements on the import statement of |
Do you have a link to a specific instance? Not seeing it in the diff on GitHub's side, but I know GitHub has been off today, so not sure if it's me or GitHub. |
Hmm I just realized those rearrangements were probably generated by @SudoSpartanDan If they were indeed generated by |
|
Yup, I didn't generate those files, so any rearranging was done by |
|
@SudoSpartanDan The If I understand the error message correctly, it's complaining |
|
Thanks, @lilincmu! Looks like that fixed the lint test! |
|
This is an awesome addition to the project thanks @SudoSpartanDan |
|
@SudoSpartanDan Thanks for adding this. We've been waiting for some sort of stateless locking mechanism for a while. Would it be possible to add Redis DB index configuration option that would issue a |
|
@Omicron7 Im not sure what that gets us for performance, also we haven’t released this yet so I want to wait til testing by others is done before we start throwing more options at this. Feel free to contribute on your side if you feel this is necessary, but in my testing, this is more of a nice to have. |
* Initial build out of redis backend * Regenerate locking mocks and matchers * More test cleanup * More test fixes * Added tests for redis * Test fix * Linting fix * Dcos update * Update redis.go * Update server.go * Adding nolint to RedisDB struct
|
Thank you so much for adding this feature @SudoSpartanDan ! @lilincmu @SudoSpartanDan how does this work with storing plans? From what I recall, Atlantis stores the plans locally on each server. Is it required for all the Atlantis instances to use the same volume so it has access to the same plan output? Have you folks been able to test this? Reason I'm asking is because i was requesting this enablement in the anton/bryant fargare Atlantis module to get more users to take advantage of this feature. The hope is that we can raise the number of Atlantis instances to more than one See request in upstream module terraform-aws-modules/terraform-aws-atlantis#322 |
|
@nitrocode we have been running this in production since it was released and have had zero problems, it works flawlessly. We don't run multiple atlantis nodes, but that I'm aware of, there's no reason why we couldn't. We perform red/black deploys using immutable ec2 infra and the atlantis user home directory is stored on EFS, so plans are persisted during deployments. |
|
That's good to hear. I'd love to know if you can increase your atlantis so it's HA. I'd also like to know if the locking db for redis also stores the plan, then we do not have to use EFS since EFS is very slow when it comes to Please let us know if you (or someone else) gets a chance to test this out. |
This is hopefully the start of many implementations of external locking databases.
This addresses #894 and has the ability to make Atlantis a stateless application within Kubernetes, allowing scaling up and down of an Atlantis instance. Additionally, this abstracts out the backend DB, making it easier for others to add other locking backend database types as time goes on.