I've the same issue when using the API with "proto_signature=".
Miles, could it be possible to increase the URL size limit in our nginx servers?
If this was an nginx problem we would get a 414 and not a 502. nginx logs confirm that the error is coming from the upstream server (the webapp). From nginx error log: recv() failed (104: Connection reset by peer) while reading response header from upstream, client: ***.***.***.***, server: crash-stats, request: "GET /search/?proto_signature=......"
Looking further, this is because of uWSGI buffer size. Increasing buffer size from default of 4096 to 16384 (out of max of 65536) was sufficient for this query to work in stage. PR here: https://github.com/mozilla/socorro/pull/3669.  http://uwsgi-docs.readthedocs.io/en/latest/Options.html#buffer-size
Commit pushed to master at https://github.com/mozilla/socorro https://github.com/mozilla/socorro/commit/507ed84a39dc3b4548aa22737898705d5ee26b83 Merge pull request #3669 from milescrabill/up-wsgi-buffer-size Bug 1338565 - webapp: increase uwsgi buffer-size 4096 (default) => 16384
This fix is now live in prod.