Sentry errors without server_name

[matthew-white]

We're seeing errors in Sentry that don't indicate a server_name, for example, this error from an ODK Cloud server. Because the error doesn't indicate things like the request URL, I'm thinking that it was thrown by the worker mechanism. I think our hope was that 9e41033 would resolve this, but it seems like these errors are still appearing in Sentry.

[yanokwa]

For the record, I'm not optimistic that #521 solves this problem.

The only tags on see on this type of error is environment, handled, level, and mechanism. I see a lot more tags on other errors (e.g., runtime, url, device).

I'm out of my depth here, but it would seem that perhaps workers don't have access to the full Sentry env? @matthew-white would it be worth it to escalate to their support team?

[matthew-white]

I also don't think that #521 solves this problem. For some of the tags you mentioned, I think it's expected that those are missing for workers, because there's no associated request (for example, url). But in other cases, it seems pretty surprising that the tag is missing (for example, runtime). I'm not sure, but it looks like this might be fixed in the latest version of the Sentry SDK: see getsentry/sentry-javascript#5190 (search for server_name).

[matthew-white]

For posterity, note that #626 effectively reverted the commit 9e41033 mentioned in the issue description.