Serverless Stack

hm I guess I only need this for non-local development where this isn't an issue

Hey <@U01MV4U2EV9> From what I’ve gathered, most folks do something like this. Per se you have a table called Posts, and you want to run change its schema. The migration steps are along the line of:
1. create a new table Posts2
2. deploy code to read from Posts, but always write to both Posts and Posts2 (so for any new items, Posts and Posts2 are in-sync)
3. run the migration lambda to fill up old content
4. deploy code again to read/write to Posts2
5. drop table Posts

And for the context of SST, if you are running `sst start`, in step 3, the migration lambda runs your local function.

I’m not sure if this is what you were asking. :grin:

When I referenced migrations I meant schema migrations in Postgres. For example if I want to add a column, I need that to happen before any new lambda code referencing it runs

So what I was hoping for was a function I could deploy first before updating any code that executes once to add the column

Think I'm able to do that with cdk custom resources but still playing with it

<@U01MV4U2EV9> I think this would be a good fit for a code pipeline stage as part of ci/cd. Locally I’m sure you could run the migration manually. Otherwise I think what <@U01JVDKASAC> said is pretty common. You could replace table with column in his example.

Let me look into codepipeline haven't used it before

I basically was trying to see if I can do everything through a lambda so I didn't have to worry about vpn/vpc access in my CI environment

I’m sure a release phase could call a lambda.

Yeah that's kind of what I was thinking earlier where I have a sst app just for my migrations that gets deployed then invoked before my main application gets deployed