Blobstore Backup Strategy Google App Engine Python

Question

I'm looking for a way to backup the blobstore on Google App Engine to recover from accidental deletion. I'd like to be able to make a backup of the datastore and blobstore at the same time, so that I could recover the entire system.

One thing I tried was using Google Cloud Storage for my blobs, and then making backup copies of buckets. However, if I delete a file in the blobstore viewer (which also deletes it in google cloud storage), and then restore the file from a bucket backup using gsutil, the blobinfo is still lost.

I believe I would have to store all the file paths (google cloud storage paths) in my datastore, and then during a restore, recreate blob keys for all those files using create_gs_key(). Is the info stored in the blobviewer (file name, content type, size, etc) stored in the datastore?

Is there anyway to backup / restore blobstore data without regenerating keys?

fejta fejta · Accepted Answer · 2013-05-18T22:32:30

Take a look at Google Cloud Storage's object versioning feature. Enable versioning on a bucket and when you delete an object it gets archived. Later you can choose to restore or permanently delete the archived version:

Create an object:

>>> PUT /bucket/obj

<<< HTTP/1.1 200 OK
<<< x-goog-generation: 12345

Then archive it:

>>> DELETE /bucket/obj

<<< HTTP/1.1 200 OK

Now it is inaccessible:

>>> GET /bucket/obj

<<< HTTP/1.1 404 NOT FOUND

However you can still request the archived version explicitly:

>>> GET /bucket/obj?generation=12345

<<< HTTP/1.1 200 OK

Until you delete the archived version:

>>> DELETE /bucket/obj?generation=12345

<<< HTTP/1.1 200 OK

Of course... this doesn't protect you from accidentally deleting the archived version.

Blobstore Backup Strategy Google App Engine Python

2 Answers