Access to logs for when webhooks are called

[pmachapman]

In Scripture Forge, we are trying to diagnose why some webhook calls are failing.

We try to fail with an error if a fault is temporary (i.e. connection), and fail silently if it is our fault (i.e. project no longer exists or similar cases).

Currently, it is hard to diagnose when a webhook call was made from Serval, and if it failed, and how many times it retried.

If possible, is there a way can we see this info, and also if possible, the message that was sent to the webhook? Maybe even Hangfire where the webhooks are called from? (in Scripture Forge we allow production access to authenticated System Admin users)

Thanks!

[johnml1135]

Let me see if I understand. There are multiple reasons why a webhook may fail (Serval or SF or network). We need some way of achieving "eventual consistency" - the way proposed is to have Serval keep track of the webhooks, and if they failed or tried and failed. Here are a few diverging ideas:

1. The question that I would have @pmachapman, is if you are going to actively query the webhooks to see if it failed (a real possibility), why not just query the engine itself?
2. The other thing I hear you saying is that you are wanting to debug why and when the webhook is failing - and by tracking the successful or failed attempts could help with that.
3. Investigate the robstness of the webhooks and raise the bar for trying to deliver them, making sure that we keep trying for at least 1 day and that we continue through server restarts.

[pmachapman]

The question that I would have @pmachapman, is if you are going to actively query the webhooks to see if it failed (a real possibility), why not just query the engine itself?

Only manually when it fails. I would not have Scripture Forge perform this query.

The other thing I hear you saying is that you are wanting to debug why and when the webhook is failing - and by tracking the successful or failed attempts could help with that.

This is the main thing I want to achieve. I guess I am am asking for is potentially more transparency on what happened for a webhook for a given translation engine when I am debugging what went wrong.