How to not lose incoming events during incidents/deploy

[Horbatenko]

Hello,

I’ve been wondering how we can ensure stable event processing when WAHA instances or services that read their data are down.

In the case of deployments, we can use a strategy that keeps at least one active WebSocket connection at all times. But how can we cover the pipeline in the event of an incident and fetch all the affected data?

After reading the docs and reviewing the available APIs, I came up with an idea for a “backfill” process, which would:

• check the timestamp of the last received message
• retrieve affected chats via /api/{session}/chats/overview by comparing last message timestamps
• fetch all messages for each chat via /api/{session}/chats/{chatId}/messages

However, this approach sounds quite complicated, so I was wondering if there’s an easier solution.

Ideally – all “unread” data would be pushed into an existing WebSocket connection so that the gap is resolved automatically.

Thanks in advance for any suggestions 🙏