Can't read snapshots when using BackupBatch

[clayrisser]

I am using BackupBatch to backup my deployments. The backups work fine.

However, when I run kubectl get snapshots I get the following error.

Error from server (InternalError): Internal error occurred: Fatal: unable to open config file: Stat: The specified key does not exist. Is there a repository at the following location? s3:s3.amazonaws.com/siliconhills-backups/stash

[clayrisser]

The stash operator also logs the following error.

/bin/restic_0.9.5 snapshots --json --quiet --no-lock --no-cache
Fatal: unable to open config file: Stat: The specified key does not exist.
Is there a repository at the following location?
s3:s3.amazonaws.com/siliconhills-backups/stash

[clayrisser]

But, you can see the backups exist.

[hossainemruz]

Good catch 👍 . We didn't update get snapshot for BackupBatch.

[clayrisser]

Ah, I see. Thanks. Here is what my repository looks like.

apiVersion: v1
items:
- apiVersion: stash.appscode.com/v1alpha1
  kind: Repository
  metadata:
    creationTimestamp: "2020-01-27T02:22:12Z"
    finalizers:
    - stash
    generation: 1
    labels:
      app: stash
      chart: stash-0.0.1
      heritage: Tiller
      io.cattle.field/appId: stash
      release: stash
    name: stash-stash
    namespace: kube-system
    resourceVersion: "45530391"
    selfLink: /apis/stash.appscode.com/v1alpha1/namespaces/kube-system/repositories/stash-stash
    uid: 92c00e01-a228-4f6a-9477-28531a9cff05
  spec:
    backend:
      s3:
        bucket: siliconhills-backups
        endpoint: s3.amazonaws.com
        prefix: stash
      storageSecretName: stash-stash-backup
kind: List
metadata:
  resourceVersion: ""
  selfLink: ""

[clayrisser]

@hossainemruz so how do I restore my backups right now? Is there any workaround?

[hossainemruz]

Use latest in snapshot field. This will restore the latest snapshot.

[clayrisser]

What's the status on this issue?

[hossainemruz]

@codejamninja In order to fix this issue, we have to fix these issues first:

• Ensure 1:1 mapping between Repository crd and actual repository for BackupBatch #1053
• Ensure order while backing up the members of the BackupBatch #1054

We will try to fix as soon as possible. However, it won't be part of v0.9.0-rc.6 which going to be out very soon.

[clayrisser]

@hossainemruz ok, thanks

[nauni77]

@codejamninja Thanks for reporting. I have the same problem.

[endrec]

@hossainemruz wrote:

Use latest in snapshot field. This will restore the latest snapshot.

It seems restoring from latest snapshot does not work with BackupBatch. 😞

I have a RestoreSession with

  target:
    ref:
      apiVersion: appcatalog.appscode.com/v1alpha1
      kind: AppBinding
      name: mongo-restore

The restore job is created with a --path flag which only contains my repository prefix (it does not contain any appbinding reference).

I can create an other repository, where I add appbinding/... to my prefix, which works wonders, but I don't think that would be the expected way to restore from a backupbatch...

So currently my working setup is:
On backup, I have a repository, with prefix backup, and a BackupBatch with two targets, a mongodb and a postgres.
(This creates snapshots in backup/appbinding/mongo and backup/appbinding/postgres.)
On restore, I have two repositories, with prefixes backup/appbinding/mongo and backup/appbinding/postgres, and two RestoreSessions .
This works, but it should be simpler, I believe.

[hossainemruz]

@codejamninja We are going to fix this issue in next release. Here are the proposed changes: stashed/project#5

[endrec]

The proposed solution looks good to me, it will make life easier. :)

[hossainemruz]

Already fixed in master.