The extract_payload part of the collector can hit issues as it's parsing the payload for dumps and annotations. Currently, we don't keep track of these issues. This is problematic because once we're done extracting data from the payload, the payload is consumed and any evidence of bugs in crash reporter clients is gone.

I want to adjust how things work so that we keep a log of how extract_payload went and include it in the collector notes. It should include the following warnings/errors:

1. when the wrong content-type (or not content-type for dumps) is specified
2. when annotation values aren't valid utf-8
3. when part names aren't valid utf-8
4. when "extra" is specified and there are also other annotations
5. when the "extra" value is malformed or the wrong type after json-decoding

Some of these are currently hard-rejects. Instead of doing that, if there's enough crash report to use, we should accept it and keep track of the errors in the collector notes.

Grabbing this to do now.

I reworked things so these are now warnings and add a crash report note and not hard-rejections:

• when the wrong content-type (or not content-type for dumps) is specified
• when annotation values aren't valid utf-8
• when "extra" is specified and there are also other annotations

The rest I'm leaving as hard-rejections for now.

Note that we need to update the crash report payload spec, too. That's in the socorro docs.

willkg merged PR #891: "bug 1760814: change some extract issues to warnings and add as notes" in c693b2c.

I deployed this just no in bug #1800935. Marking as FIXED.