Is it possible to catch a handler that is down and continue generating the schema without it?

[Davenporten]

I'm currently experimenting with using graphql-mesh to bring together a large number of services across a platform. Due to the fact that many of these services are new and still in development in many ways, but still need to be integrated with our mesh, they are sometimes unavailable when our graphql-mesh instance is being built. When this happens the whole build process fails and we either have to wait until the service it back up or release a version without it. Ideally we'd like to do both, meaning, we'd like to release without it, but not have to explicitly remove it, so all we'd need to do is rebuild the instance once the service is up.

Is there a way to do this? I understand why this wouldn't be the default functionality, but is there some configuration that would allow the build process to continue without the downed service?

As always, thanks so much in advance!

[dr3]

+1 Weve had the same problem & have discussed having backups in git or s3 which it can fallback to, or like you say, skip it entirely for local dev

[Davenporten]

I'm kind of wondering if I could create a custom introspection for the handlers that could test to see if the service is up.

[dr3]

It also would be good to have better logging in openapi mesh for when a source is down. We have swagger endpoints which sometimes return a generic html 404 page and it gives errors like Unexpected character <