Jon Atkinson | 2024 Week #38 Notes

2024 Week #38 Notes

week-notes

2024-09-16

Early start.
Moved ahead this PR for JQ
Planning for Tailscale SSH migration later today.
- live2 - old IP: 100.xxx..xx3.46 - new IP: 100.xxx.xxx.52 - ext IP: 157.xxx.xxx.156
- dev - old IP: 100.xxx.xx7.46 - new IP: 100.xxx.xxx.116 - ext IP: 116.xxx.xxx.xxx
- internal - old IP: - new IP: 100.70.179.31 - ext IP: 159.xxx.xxx.143
$ sudo ufw status numbered $ sudo ufw allow 22/tcp # Allow from all interfaces while testing. $ sudo tailscale down $ sudo tailscale up --ssh --advertise-tags=tag:server $ sudo ufw status numbered $ sudo ufw allow in on tailscale0 to any port 22
Spoke with WH regarding in-deploy backups.
Merged this small documentation change from last week.
Completed the Tailscale SSH migrations. There will need to be work done on live3 when it's ready.
Debugged some Tailscale deploy problems post-SSH migration.
GG marketing planning with VP.
SWNS and PF meeting with MB.
Spent an hour doing the cybersecurity questionnaire for PF.
Ruminated on whether to roll back the SSH changes or to just plough forward.

Early start, helping JQ with a manual deploy.
Wrote docs for manual deployments.
Caught up with RT about plans for change to deploy methods.
Wrote changes in github-actions to build containers, this branch, unmerged.
Fixed certificate error for giant-checklists.
Worked with R to get to a successful build of a container and push to GHCR.
Further work with R to get Tailscale SSH in place on github-runner. Did something stupid with tailscale down, and managed to lock myself out.
- Setup rescue console.
- Mount chroot in rescue image.
- Manually amend the ufw rules to re-enable access from the public interface.
- Reboot and back in.
Reconfigured Tailscale SSH tags and ACLs to add a runner role.
Reconfigured github-runner to disable the public SSH access again.
Tried to get some Teams deployment reporting working with this package, but it seems that the documentation doesn't match the code. Found some updated versions in the Network, going to try those instead. R offered to help.
Late discussion with JM/SWNS about providing database backups. Unclear still.

Chatted with R about the deployments. We need to solve the image tags being considered as 'already deployed', even though they might have changed. Dokku is quite naive, and only checks the tag name, rather than the contents of the image.
Wrote up conculsion for the VP/A refactor proposal notes. L told me my prose sounded like ChatGPT, what a savage burn.
1-2-1 with RM, chatted about databases and role0-based access control.
Discussion with chatbot about a minimal Django example, in a single file. Results needed prompting, but reasonable. Put out this article about it.
1-2-1 with KS. Discussed some data visualisation stuff.
Added syntax highlighting to personal blog. Also pushed commits to handle 404 and 403 pages, which I previously implemented and deployed but never committed.
Chat with GF re: Giant Carbon dashboard and some visualisation work.
Wrote context_processor for product updates feature for GG.
- Quite a loaded name, "updates".
- Is User.last_product_update_guid the right place for this?
- Review caching.
- Got this mainly working, going to pair tomorrow with SP to finish it up.
Reviewed RT's work on deployments, looks good.
Opened this draft PR for the work on product update notifications.
Did some work with AI summarisers to build environment variable documentation for GG.

Booked travel for Alumnites on 16th Oct.
Spent some time looking through the Sentry and Dependabot docs.
Reviewed WH's H4H pull request.
Merged this PR from last night for the GG docs.
Useful rubbed-duck with SP to fix the product update notifications PR.
Further fix for that PR to make it available in the docs navigation.
Finished this PR for the product updates feature in GG.
Briefed JQ for frontend.
Battled for 15 minutes formatting a USB drive for a BIOS update.
Worked some more on giant-carbon, implemented a pure-python page weight calculator. It's not as accurate as the Lighthouse-based solution, but it's withing 90% of the Lighthouse calculations, so it's good enough.
Had a chat with TB about page weight calculations, and a bug I found while testing.
Merged the GG product updates PR to develop.
Did some work on the deployment notifications, tested on internal, which worked 'okay'. They're not the most ergonomic but they are there, and that's enough.
BIOS update on main computer.
Review GG dashboard for ES

Forecast SSO reconfiguration.
- Documentation very outdated.
- Sent bug report to Forecast detailing problems with the setup experience.
- Updated staff.
JQ delivered frontend for GG product updates feature.
Discussion with TB again about video elements on the web and their autoplay status. To figure out what should be included in my page weight calculations.
Updated deployment notifications for actions.
- Explored the ActiveCard layout stuff.
- Am I going to need to write my own custom Github action for this?
Write new Github action, here, which encapsulates the Teams notification code.
Updated github-actions.git to use this.
Test deploy. Successful.
Further updates to github-actions.git to deal with failures more elegantly.