Issue calling CloudProviderSnapshotBackupPolicies Service

[makubit]

Hi Team!

I'm trying to call Atlas API to GET the backup policy (CloudProviderSnapshotBackupPoliciesService) and I'm getting an error from the client:

{"responseError": "json: unsupported type: func() (io.ReadCloser, error)"}

My cluster is running in Atlas, I have enabled continuous backups. I'm passing valid parameters and it seems that issue is not on my service side.
Could someone look at this?

Cheers,
Magda

[gssbzn]

Hi @makubit can you please provide a sample application that reproduces the error?

[makubit]

Sure, the code is pretty easy, just call the mentioned service:

	t := digest.NewTransport(string(publicKey), string(privateKey))
	tc, err := t.Client()
	if err != nil {
		return nil, fmt.Errorf("error creating http client: %w", err)
	}
	mcli := mongodbatlas.NewClient(tc)

	ps, resp, err := mcli.CloudProviderSnapshotBackupPolicies.Get(ctx, groupID, clusterName)
	//here returned an error

Btw It would be easier to understand this service if there were named parameters in the code:

type CloudProviderSnapshotBackupPoliciesService interface {
	Get(context.Context, string, string) (*CloudProviderSnapshotBackupPolicy, *Response, error)
	Update(context.Context, string, string, *CloudProviderSnapshotBackupPolicy) (*CloudProviderSnapshotBackupPolicy, *Response, error)
	Delete(context.Context, string, string) (*CloudProviderSnapshotBackupPolicy, *Response, error)
}

Get(context.Context, string, string) doesn't show what that strings mean, it's just guessing the convention.

[gssbzn]

Hi @makubit I took your example to make it complete and I'm not able to reproduce

package main

import (
	"context"
	"fmt"

	"github.com/mongodb-forks/digest"
	"go.mongodb.org/atlas/mongodbatlas"
)

func main() {
	t := digest.NewTransport("public-key", "private-key")
	tc, err := t.Client()
	if err != nil {
		panic(err)
	}
	mcli, _ := mongodbatlas.NewClient(tc)
	ps, _, err := mcli.CloudProviderSnapshotBackupPolicies.Get(context.TODO(), "project-id", "cluster-name")
	if err != nil {
		panic(err)
	}
	fmt.Println(ps) // works
}

Btw It would be easier to understand this service if there were named parameters

Feel free to open a PR to improve this

[gssbzn]

I also tested this via the CLI and

atlas backups schedule describe <clusterName> -o json

also works as expected

This feels like it may be a configuration issue the client on your side and why the closer your example is to how you are using it it would be better for us to diagnose

[makubit]

@gssbzn Thanks for checking! I noticed that one of my conditions was wrong - I didn't wait for the finishing of cluster provisioning and that was the issue (I didn't notice as the message doesn't directly say what was the issue).

In my free time, I can prepare PR with named parameters for a better understanding of the service.

Sorry for the false alarm.

[gssbzn]

I tried the following on the CLI (mainly because it's easier to create a cluster)

atlas cluster create M10 --provider AWS --region US_EAST_1 --tier M10 --backup
atlas backups schedule describe M10 -o json

And the error I get is

404 (request "BACKUP_CONFIG_NOT_FOUND") No backup configuration exists for cluster 63d3d71bbc21f7794cc0270d in group 5e429f2e06822c6eac4d5d9.

And for a cluster that doesn't exists I get

404 (request "CLUSTER_NAME_NOT_FOUND") No cluster with name "fake" exists in group 5e429f2e06822c6eac4d59d9.

I'm a bit curious because this error in particular

unsupported type: func() (io.ReadCloser, error)

Seems like a deserialization error somewhere and we have some options in the client to customize where deserialization happens instead of a return value.

If it works for you now that you wait for the cluster that's great but if it happens again let us know with a way to reproduce which is as close as how you are using the lib so we can discard any internal issue with some of the more advanced settings