Backup vs third party clients

Do you have any feeling for how third party apps will handle the rollback to a backup?

In particular, are there any sequential ID numbers, or are you using GUIDs or similar? In other words, if I mark an article as read while NewsBlur is offline and then NewsBlur comes back and loads articles in a different order (or is missing one, or whatever) could I end up marking the wrong article read or otherwise getting out of sync?

I’m primarily using Reeder for iOS.

Also, best of luck with the recovery!

Everything’s under GUIDs and the backup is not actually a backup but rather the old primary MongoDB machine taken off the cluster. So things should look identical to a few hours ago.

Right, but let’s assume my clients updated more recently than that. And let’s also assume one or more of my feeds has items that were published and have now been removed (likely, in my particular case).

So my clients might have articles 10, 11, 12, 13 and 24, the database restore will have up to 10, and will now pull 11 and 13, but it’ll never see 12. What will happen to my client?

In the case of IMAP this would be permanently out of sync unless the server rotated the UIDVALIDTY field so that the client knows to not trust the local cache and to pull a full free update from the server.

In the above example the IMAP server would present 10, 11, 12, but what it calls 12 is what my client would call 13, which is where the out-of-sync state happens.

What I’m not sure about is how NewsBlur presents stuff to clients. If it is by GUIDs or IDs from the feeds then everything will mostly just work, but if it is using incrementing numbers then I’ll need to wipe my clients and reconnect them, unless you have a way to handle this from the server.

Sorry for the hassle, just want to make sure as Reeder has some “mark article as read” actions queued already and I don’t want to lose some articles by having the wrong ones get marked as read.

1 Like