Rethink the logging of BridgeDB
Now that we are playing around with BridgeDB, it's probably a good idea to also improve its logging.
For example, BridgeDB currently doesn't have any debug logs during descriptor parsing (which made it harder to find the root of #9174 (moved)).
Also, we should make sure that BridgeDB can scrub client's IPs from the logs (even for debug logs). Example of offending logs: Valid recaptcha from ... and [DEBUG] getBridgesForIP(, ...)
Also, maybe we should consider using Tor's extended nicknames to improve log greppability. Extended nicknames combine the identity digest and the nickname of a router in the same line. Example:
$AAAAAAAA01234AAAAAAAAAAAAAAAAAAAAAAAAAAA=fred
.
Check Tor's function node_get_verbose_nickname()
and https://gitweb.torproject.org/torspec.git/blob/HEAD:/control-spec.txt#l152
We should also identify other information that the BridgeDB operator would be interested in and log them (maybe even write a heartbeat log with statistics).