Have webhooks become flaky?

I have a service which sits idle on Heroku waiting for webhook events from MRs, and performs an action whenever an MR is closed.

Over the last few weeks, this service (which was previously reliable) has become flaky, and seems to run on some MR close events, but not all. There doesn’t seem to be any pattern to which MR close events correctly fire the bot and which do not.

Has anyone else had problems with webhooks lately?

I have a problem with one of my webhook: it’s giving an SSL errror certificate expired when the cert is perfectly fine on the target server. Do you see an ssl error in the logs, too?

@Nic0 I don’t have SSL certs on my webhook endpoints, so I think you and I are probably seeing different issues.

Hi, we are also experiencing problems with webhooks.
We have a Jenkins instance which runs Multibranch jobs whenever MR opens/closes.
Over the last few weeks we noticed significant delay between the time when MR is actually opened/closed in Gitlab and relevant Jenkins job is triggered.
For instance, this delay causes that MR can be merged, although Jenkins job for this MR will eventually fail (we configured that every MR must have successful pipeline).
This is because Gitlab will trigger Jenkins job when new branch is created (the one that participates in MR), in which we do not do anything and job finishes with success. Immediately after MR is created, merge button is green and clickable, although relevant Jenkins job for opened MR is not started at all. Only after >10s Jenkins receives payload, runs the job which blocks the MR to be closed.
We are constantly updating Gitlab instance and this behavior started to manifest after migration to 15.0.

I mean, we could change the definition of Jenkins job to not finish with success when not one stage is executed, but I do not think this should be the road we should take.

@asoldo11 That’s interesting. I’ve had a look through the webhook bugs and I can’t find anything specific to flaky webhooks.

yes, me too. i will look into it a bit deeper and try to find out what might cause such strange behavior.

Thanks @asoldo11

I’ve raised an issue here - maybe you can upvote it and add any useful information you can gather?