Log session ID instead of client IP during captcha flow

[louisgregg]

Logging the IP address of a client during the captcha registration flow is a privacy issue. Client IPs are stored in the database along with timestamp and UserAgent, so it seems redundant to log them here, where it would may be easier for an attacker to extract them, depending on how the Postgres DB is protected.

Therefore, I have modified the code to log the user session instead, which isn't PII at least. I'm not entirely sure how sensitive these session IDs are in the context of this application, or in the captcha flow. Does logging the session ID pose a security risk? Please let me know - I am happy to modify the PR and tests to log neither session ID nor IP address instead.
Thank you

Pull Request Checklist

• Pull request is based on the develop branch
• Pull request includes a changelog file. The entry should:
  • Be a short description of your change which makes sense to users. "Fixed a bug that prevented receiving messages from other servers." instead of "Moved X method from EventStore to EventWorkerStore.".
  • Use markdown where necessary, mostly for code blocks.
  • End with either a period (.) or an exclamation mark (!).
  • Start with a capital letter.
  • Feel free to credit yourself, by adding a sentence "Contributed by @github_username." or "Contributed by [Your Name]." to the end of the entry.
• Code style is correct
  (run the linters)

[CLAassistant]

All committers have signed the CLA.

[louisgregg]

How can I effectively test against the trial-olddeps setup in your pipeline? This build step is failing
https://github.com/element-hq/synapse/actions/runs/11988497284/job/33428556861
I believe this is because I am developing against Twisted 24.7.0 and the trial-olddeps step uses Twisted 18.9.0. I am going to try to replicate the build steps on my machine by running the prepare_old_deps.sh script. Otherwise, do you have any developer documentation on how to get this environment set up correctly?
Thanks!

[reivilibre]

I can certainly see that logging the client's IP in this line is redundant, because that should already be getting logged for the request's log line that Synapse logs for every request.

I could get behind just removing the client IP from this log line and even downgrading it to debug instead of info, since it seems a bit needless to just log the input from the client.
I don't personally see much benefit in logging the session for this flow on its own, so we could leave that off too.

However, I feel like I should mention the Synapse logs still contain more PII — User IDs, IP addresses on the request lines, etc — that I don't feel this PR really makes much difference on that particular front?