emails through democracy.io are returning various error messages

[glorioso]

We are using a fork of democracy.io to send out emails through phantom of the capitol. We are noticing that a large percentage (30-50%) of these are returning various error messages. Is this a known issue?

Here is our tool:
https://github.com/AcademicsForTheFutureOfScience/afs_contact_congress_website

[j-ro]

You'll need to be more specific, but there are no general known issues besides the ones you already see in the issues section.

[b-carter]

Stepping in to provide some more details on the issue our site is having that @glorioso mentioned. I have a log of the responses from POTC that are given to democracy.io.

One of the ones that has an error status reads:
[{"bioguideId":"G000555","status":"error","uid":"ba62be9d3189e5480d5914bc834794a0"}]

(Most of the responses return "success" status but a decent percentage (seemingly random) return in error.)

We're using the EFF hosted POTC server at https://congressforms.eff.org. Are you able to look into your logs using that uid to get more details of the exact error?

[j-ro]

You'd have to talk to EFF about that, I'm not them, and this is the public project, not their instance. If you have a debug key, the other endpoints should give you more detail.

Generally, a few errors are normal, though. Forms can error for reasons beyond issues with the server or with a YAML file. For example, some forms only take certain lengths of input, and if folks type more for their form fill, it'll not be successful, and thus error. Similarly, sometimes filtering in Congress on the inputs will flag some responses as possible hacking attempts (like if you have text that looks like code in the response). That kind of thing.

So, if you're seeing continual errors on a certain bioid/YAML, it means likely the YAML is bad -- with detail you can fix it and contribute to the repo here: https://github.com/unitedstates/contact-congress

But if there's intermittent errors, that's likely somewhat normal operation. In our implementation, we get 90%+ successful fills, but not quite 100%.