All errors are reported in development mode, but on staging and production almost all errors are ignored despite seeing DEBUG 2016-11-07 19:32:10,623 base 2453 188356840266813 Sending message of length 3187 to https://sentry.io/api/000000/store/ in the log
Quick suggestions:
- Verify that you haven’t blocked your production host in Project Settings (see Allowed Domains)
- Verify that you have specified the right DSN in production
If you are observing an outbound HTTP request (that returns 200), that almost certainly means Sentry received the data.
almost all errors are ignored
Does that mean that some errors are not ignored?
benvinegar,
Yes, only a very, very small number (greater than zero) would show up, seemingly regardless of which error it was or which system it came from. I don’t see the problem any more, so maybe it was an error at sentry.io? I did see a few times URLError: <urlopen error EOF occurred in violation of protocol (_ssl.c:590)> but that error was not reproducible on-demand either. It almost seemed like a race condition or something. enable-threads is set, as well as re-fork and refork-post-jail. I got nowhere on this, despite spending all day on it. I was ready to remove sentry completely when the issue disappeared.
Re: "observing an outbound HTTP request (that returns 200)"
Sentry/Raven does not log the return status code at all, that was one of the many big problems. The log message is above in the original question, with numbers changed. So, troubleshooting tips on this are always welcome.
b264