Tail worker stopped working between deploys (~a week apart) without any code changes
1) I have a worker that is using itself as a tail worker. Call it
worker-a
2) the worker with that setup was initially deployed ~2 weeks ago and then a few more times with code changes over the next week (up to ~1 week ago)
3) those deploys sent logs to datadog for ~2 weeks but stopped when we deployed yesterday.
4) The only difference between the working week old code and yesterday was the version number for our release inside a package.json. In other words, no dependencies change, no library or app code changed.
5) A different worker, worker-b, was using worker-a as its Tail Worker and it continued to sends logs to datadog for almost a day after worker-a stopped logging. worker-b only stopped when it's week-old deployment was replaced. In this case, it was identical code to the working code. Literally just wrangler deployed the same SHA again
The workers otherwise seem to behave as expected. There's just zero logs coming into datadog and the gaps start at the same time as the worker deployments.
The (total?) lack observability for tail workers makes this extra challenging. I don't know if code inside the tail worker is failing or otherwise logging helpful info. Just that logs are down to zero.
Are there any theories on what the issue could be? Any thing I could do to further debug and/or narrow it down?
Can I give someone here my account id and some worker names and have them look into it, or is sending a "usually takes three days" support ticket the only option?0 Replies