yuth
commented
4 years ago Most of this comes directly from an Aurora instance, however a growing amount come from other APIs such as lambda functions.
Are these functions provisioned using Amplify CLI? If these are provisioned using Amplify CLI, we expose an environment variable called ENV
To add one of these is not a huge deal, as we can add it to the JSON file for the datasources/functions/resolvers, however the issue is that these values are all unique per environment.
Where do add the JSON file? Is it in the lambda function itself or are you switching these using a custom stack?
bduff9
stale[bot]
kaustavghosh06
github-actions[bot]
Is your feature request related to a problem? Please describe. I apologize if this already exists (and would be thrilled if it does) but could not find an answer after looking in the docs and past issues. We have a large production application that uses AppSync for all data-related needs. Most of this comes directly from an Aurora instance, however a growing amount come from other APIs such as lambda functions. To add one of these is not a huge deal, as we can add it to the JSON file for the datasources/functions/resolvers, however the issue is that these values are all unique per environment. Currently, this poses 2 challenges. First, when adding a datasource and merging to a main branch (i.e. develop to qa), we have to merge the new code into the environment, and then manually change the datasource info, commit this and then push it up using
amplify push. Not a big deal but being a manual step, would be easy to miss. Second, we have our datasources for lambdas defined, and then in each resolver there is a path component that points to the current environment that also needs to be manually changed (i.e.resourcePath: '/dev/rest/of/path',toresourcePath: '/qa/rest/of/path',). This is also a manual step like the first part but is a bigger deal since future changes to this file could cause overwrites/conflicts.Describe the solution you'd like Unsure if there is a current way to manage this, but some way to add the right datasources per environment, and then not lock the resourcePaths to the current environment to make deploys much easier/less prone to breaking.
Describe alternatives you've considered As mentioned, this has been working for a month or two, but is too manual and can cause hard to track issues (like it just did earlier today).
Additional context We might not be following the best practices here, so if there is a way to make this better/easier, I am open to that as well. Just could not find the answer after looking so hoping one of the experts here can point us in the right direction.