Team:Sysadmin/Lessons Learned

From WHY2025 wiki
Jump to navigation Jump to search

Lessons learned

Discussed on 2025-09-08

General

  • Bad: Insufficient resources (CPU/RAM). Consider a dedicated server again for next event for the more heavy services.
  • Bad: No budget requested.
  • Bad: Orga is extremely bad at planning ahead in terms of requesting resources/services. Many things had to be configured after the communicated cut-off date.
  • Good: Onsite server worked perfectly fine, even though it was requested after the cut-off date and we had to improvise.
  • Improve: Better document procedures and/or create standardized forms.
  • Bad: As mentioned last event: power can and will fail. *Definitely* get a UPS?
  • Bad: Be more proactive in communication w/ other teams.
  • Bad: Other orga seems harder to reach out to.
  • Improve: Make a timeline of when to bring up services during WHY+1 early start of team. Document this on the wiki.
  • Bad: Setting/Fixing up specifically requested things. Some topic had a tendency to stick around for multiple meetings w/ no progress in sight.
  • Good: VPN's for Terrain DB worked for everyone as far as we know. Collaborative real-time map editing appears to have worked quite nicely.
  • Bad: Some people apparently needed multiple certificates (concurrent connections), but wasn't mentioned to us till we were on the field (far too late). Breakdown in communication?
  • Improve: Consider the bureaucratic/workload impact of security measures.
  • Improve: figure out a better way to vet VPN requests (cut down on communication, delays).
  • Improve: Figure out an LLM-policy for orga teams
  • Improve: Further document internal procedures.
  • Improve: Have a take-in list for all teams during the first WHY+1 orga meet.
  • Good: A lot of memories (snaps) were collected during the event. This should prove useful for the website and Team:Info's booklet/promo next time.
  • Improve: Visibility of the team to other orga.
  • Improve: Get a flex working spot next time around. Yes, this costs money. It's worth it.
  • Improve: Don't be extremely tight on finances/costs, see prior point for example.

Service: collabora

  • Bad: had tons of issues during use by other orga: black screens during initial connect, broken documents, prints missing data, copy-paste not working, etc etc
  • Bad: this didn't seem resource-bound, machine had plenty to spare
  • Do not use this again next time.

Service: Grafana/MQTT

  • Good: More data points than ever
  • Good: More things integrated (consumers) into the feed.
  • Good: More data more better.

Service: angelsystem

  • Bad: barrier of responsibility needs to be made more clear.

Service: authentik

  • Improve: Next time, set up the authentication lifetime (remember me) up way earlier...
  • Good: No complaints from our side otherwise

Service: Everything Kanban related

  • Improve: Figure out a better selection of service to host next edition. We had 3 separate options running at the end of the event.
  • Improve: Figure out orga requirements for such software.
  • Live updating kanban boards are a must, otherwise this will lead to data corruption in some cases, and it makes meetings more chaotic if everyone has to refresh the page all the time.

Service: Support queue

  • Bad: No inter-team support ticket
  • Bad: uses exorbitant amount of resources
  • Bad: unintuitive UI design, not allowed to open multiple tabs
  • Good: reliable, few issues once it was up and running

Service: Tickets

  • Bad: Not all organizational constraints were known to us ahead of time.

Service: Wiki

  • Bad: People appear to be less technical inclined. Wiki's are seen as too challenging to update.
  • Improve: Enable the visual editor next time.