Staff cannot dismiss notifications due to Babble plugin

Hello! We recently upgraded our self-hosted deployment to the latest and greatest Discourse (2.9.0.beta1, 5a93ce421d), but we’re observing a huge increase in staff notifications (compared to before the upgrade, from 2.8), and we haven’t been able to figure out how to clear any of these notifications from our Staff notifications.

I have dozens of notifications from the last few hours of activity, but clicking on the “Dismiss” buttons in either the notifications dropdown or the dedicated notifications page in the profile doesn’t appear to do anything - all of the unread notifications remain unread:

I’m not sure why some notifications are marked as read while others aren’t - I have opened each of these individually to investigate, but their read/unread status doesn’t seem to change. As best I can tell this is only an issue with Staff, but that is unconfirmed.

Can anyone point me in the right direction for resolving this? I noticed that the network tab shows some service worker interactions when clicking “Dismiss” - do I have a stale service worker? Or perhaps an old admin setting with new behavior?

Thanks so much!

Scratch that - all of our user accounts appear to have a huge amount of notifications that aren’t relevant to them, including staff notifications, and I also cannot clear those notifications. I checked in a few different browsers, too.

EDIT: it almost looks like all users have one shared set of notifications - the reports that are coming in all seem to indicate the same set of notifications showing up in everyone’s notifications panel.

open and see if there are any that are stuck

delete the items in sidekiq/queues if you must

1 Like

Thanks @IAmGav - the retries page says “No retries were found” which sounds like that is all good to go.

Delete the actual queues? Will I need to recreate them manually?

This is what it should look like

If you delete it, it will recreate it.
It is to stop spamming ppl with notifications

1 Like

Thanks, I removed the queues, looks like they have reinitialized themselves now, which is great.

I can’t quite tell if your problem is resolved.

It may be that redis was down for a while and you’re only just now getting all of those notifications.

Another possibility is that you have a plugin that’s causing you to get too many notifications. I remember seeing something about that in the past but can’t remember what plugin it was.


the indexes are corrupt and need to be fixed.


I believe it was babble:

1 Like

Hi all, thanks for the follow-up - @IAmGav was helping me in a DM.

Yes, the indexes are an issue - we were unable to restore from a recent backup because of this. We removed duplicate records from a backup (manually) and were able to successfully restore (using the built-in discourse restore cli tool).

The “shared global notifications” are still present - we’re removing the Babble plugin now, will respond when we have more info. The silly thing is we weren’t using the Babble plugin, it was installed but not enabled.

Thanks all for your help!


Hey folks, removing the babble plugin seems to have resolved the “shared global notifications” issue for us - thanks for all your help and suggestions.