Closed slackermorris closed 5 months ago
Here is a resource from Github: https://github.blog/2022-02-02-build-ci-cd-pipeline-github-actions-four-steps/
I need to set up CI/CD pipeline for deploying my SST Stack.
I need to rename the stage. staging
is essentially called slackermorris
.
I wonder where the envs
are stored? I think I really need to read this document to better understand the full stack experience: https://docs.sst.dev/constructs/StaticSite#environment-variables
master
.I suppose another thing to be done is to point routing to my application and buy a domain.
OK. Sort of unrelated. But I requested a new domain for badreviewsmakegoodmovies
.
Seems like I need to renew ownership of my other domain.
I have made a request for the domain.
OK. Now I am trying to make use of this domain
. I am reading the source here: https://sst.dev/chapters/custom-domains-in-serverless-apis.html
This is helpful to see what the config actually looks like: https://github.com/sst/sst/issues/3139
I am in a bit of bind. I thought I deleted all the CloudFormation Resources but it doesn't seem like it. And now I am stuck with some weird UPDATE_ROLLBACK_COMPLETE_CLEANUP_IN_PROGRESS
error.
I triggered a deletion from the AWS Console.
I need to delete all the prod
not production
Stacks. They were misdeployed.
Cool. I am deploying and then I will have to update the endpoints in the client app. It seems like the Site
Stack is hanging. I think I have hit this before.
I think I will let it run for a while longer and if nothing happens then I'll trigger a deploy.
Whoops. Wasn't meant to do that. Lol.
I am deleting the production
Site Stack from CLI.
I wonder if there is a step in particular that it is hanging on?
This might be what I am after.
This might be entirely unrelated.
Found here.
I mean, I feel like I have things set up correctly.
I triggered deletion via the AWS Console again. I will then try to deploy it.
Here is a bit more information.
So, looking at the log groups
for the different lambdas. I can find a certificate lambda that seems to have a log entry for the deployment period.
It looks like it depends on aws-sdk
and I don't have this as a explicit dependency of my project.
Though sst
does seem to "bring it in".
I am trying to see if I get the same issue when deploying to my development stage. If it shits itself then I will add aws-sdk
as a dep.
So, deploying to development worked fine.
I am trying to delete the production
Stack once more.
In the meantime, maybe it is a good idea to have an alias for the development | staging
API?
There is something buggy about this, because this is the only difference between production
and staging
.
Remember that the ApiStack
deployed just fine.
Maybe I remove that code and see if I can deploy.
I am trying to deploy to production
without the domainAlias
mentioned. I think I am still running into the same issue. It looks like it is hanging here.
This is the problematic lambda.
I am really not sure if it is as simple as this, but it might be. A lot of my application code lambdas are deployed with a relevant package.json
nearby.
It begs the question: why does staging
deploy just fine?
This looks like the difference in deployed lambdas between the environments.
Let's see what happens if I specify the domain for staging
as well.
I bumped SST dep as per the recommendation in Discord.
Maybe I didn't bump high enough?
It's still hanging.
It's still hanging here.
Just upgraded to the latest version of sst
which includes bumps in dependencies.
I tried to re-run the production deployment but I've run up against the same thing:
I guess I have to wait until CloudFormation stops the whole process, etc.
I started a delete from the AWS Console.
Alright giving it another go with updated dependencies. It seems like it is hanging in the same place.
There might be a little bit of movement.
It fucking worked. Holy shit.
I can probably merge all the stacks now.
What else do I need to do to be production ready? I think having a CI/CD pipeline set up would be hugely beneficial.
Testing is good. Features can be added.
Maybe it is best to deploy to production and buy a domain?