Sprint 1.44.0 2/2 - Jan 23 to Feb 3

[lharries]

Global Sprint Planning

3 things that might take us down

1. Rate limits, had an issue with local evaluation of feature flags being hit. Team experimentation picking it up
2. Santize sub queries. Marius/Yakko on it
3. Recordings table is growing. Short-term fix can be make the tables bigger. Need to find a longer-term fix too. Team infra will pick up short-term fix next sprint as part of infra hero. Ben will evaluate a longer-term fix
4. Keys not having enough values. @fuziontech will review the urgency.

Retro: Status of Outcomes from Previous Sprint

1. Data Exploration - get events table live to real users. Get pivot tables out for internal testing. Finish the last insight types. @mariusandra Why? Leftover from last time, blocker for PH3000, people want to explore data in ways we don't support. --> Good progress. Events table tested with real users.
2. Break down PostHog 3000 into next steps, and solve prerequisites (antd). @Twixes Why? PostHog 3000 is our Q1 OKR, and required to nail slick mode. To create an interface better suited for power users. --> Work continues. Loads of ants.
3. Performance: Property definitions, filters and data types @macobo Why? Event property definitions load slow, and person property definitions lack type information (e.g. string vs datetime), making some queries impossible for a few users. --> Karl on support & ill, now making progress, will partially carry over into "slick filters"
4. Internally test and then finish up Unified Network panel to gradually rollout to customers @alexkim205
   why? - This rework brings lots of 3000-esque improvements and is a pre-cursor to Performance panel.
   • Decided on what it will look like. Now in the process of launching it
5. iOS Recording groundwork - project setup, working example recording to file @benjackwhite
   why? - Get a head start on our hardest goal and answer the unknown questions
   • Ground work is done ✅
6. Hook up pricing pages to billing service for faster pricing iterations @kappa90
   why? So we can show different pricing to different cohorts and have faster pricing iteration
   • YAML method done and works for in-product. Website integrating rolling over
7. @ellie: Productionize US postgres replica, explore cutting over to it
   why? Unify infrastructure
   • Spent time hiring instead - offer out! Will do next week
8. @ellie: Work with @kappa90 to improve scheduled billing jobs
   why? Allow billing to run scheduled task with more visibililty/reliability
   • Spent time hiring instead - offer out! Will do next week
9. @danielxnj: NAT gateway per AZ, saves $$$
   why? save $$
   • Done ✅
10. @danielxnj: Clickhouse backups in S3 to save $$$
    why? Save $$
    • Still ongoing work, tricker than expected
11. Feature Flag resiliency https://github.com/PostHog/meta/pull/74/files @neilkakkar
    why? If our DB is down, feature flags still return something
    • It's live but haven't tested it yet. Need to update client libraries in the next sprint.
12. JSON Flags, allow users to configure flag responses Feature flag support for Strings #13144 @EDsCODE
    why? People want to use feature flags for more things. Requested a lot.
    • Backend completed, finishing up the frontend now
13. Offsite @fuziontech
    why? Make sure we carry momentum from offsite.
    • Came up with many PoE solutions and executing on it now

Retro: What can we do better next sprint?

Support hero this sprint

Week 1: Paul H
Week 2: James G

Plan: Proposed Goals for Next Sprint

Each goal should have a single owner. Owner can only be an engineer.

1. Hook up pricing page to billing server @raquelmsmith (outstanding)
2. Rate limiting when customers are beyond payment limits with special guest @benjackwhite & @kappa90
3. Defaulting users into a paid plan on signup @kappa90
4. Data exploration: tiles on dashboards @pauldambra
   • Why? Last big piece in HogQL & Data Exploration next steps meta#81 to tie the unified vision together.
5. Data exploration: slick filters @mariusandra
   • Why? HogQL is powerful and soon feat(data-exploration): HogQL property filters #13264. Make the HogQL editor slick.
6. PostHog 3000 prerequisites: continue yeeting ants @Twixes
   • Why? Q1 OKR. Also: we ship about 800kb of CSS. About 500kb of that is antd. Not slick.
7. Rate Limiting all API endpoints (slack thread) @neilkakkar
8. All other client and server libraries for resiliency and JSON payload @EDsCODE
9. @guidoiaquinti: Finish up ARM usage, scale up the usage and make sure nothing breaks.
10. @ellie: Continuing working on postgres! Didn't quite go to plan last time
11. PoE “Person Overrides” ingestion shipped and queries enabled for team 2
    • why: Query Performance while maintaining Data Quality
12. Rollout Network panel to some beta users @alexkim205
    • why? We have been running it successfully internally last sprint and the final touches are there so lets get it out.
13. Turn on Recordings by default @alexkim205
    • why? Huge number of customers we are missing out on as they don't realise it isn't on by default.
14. Mobile Session Recordings published (alpha)
    • why? Makes sense to test what we have ASAP with real apps (e.g. PostHoger's apps)

[kappa90]

Team Growth 🌱

Retro

• We created a biweekly pairing session, which is extremely helpful
• Achieved a new Plans API together
• Finished migrating billing v1
• Pricing principles & pricing experiments RFCs
• Deprecated self-hosted billing on the billing service
• Fixed flickering billing warnings and Quick Start button

Hang over items from previous sprint

• Verify email addresses @raquelmsmith
• Rate limiting (we decided to drop the current approach and restart together with pipeline and recordings folks) @kappa90
• Hook up billing tables to plans API (in-product one will be done by end of current sprint) @raquelmsmith
• Deprecate the license server @kappa90

Planning

High priority

• Hook up pricing page to billing server @raquelmsmith (outstanding)
• Rate limiting with special guest @benjackwhite & @kappa90
• Defaulting users into a paid plan on signup @kappa90

Bugs

• Nope

Low priority / side quests

• Verify email addresses @raquelmsmith
• Billing emails @kappa90
• Improve onboarding flow to explain better session recordings and autocapture @raquelmsmith

[mariusandra]

Team Product Analytics next sprint

(will bewas moved)

Retro

• Data Exploration - get events table live to real users. Get pivot tables out for internal testing. Finish the last insight types. @mariusandra Why? Leftover from last time, blocker for PH3000, people want to explore data in ways we don't support. --> Good progress. Events table tested with real users.
• Break down PostHog 3000 into next steps, and solve prerequisites (antd). @Twixes Why? PostHog 3000 is our Q1 OKR, and required to nail slick mode. To create an interface better suited for power users. --> Work continues. Loads of ants.
• Performance: Property definitions, filters and data types @macobo Why? Event property definitions load slow, and person property definitions lack type information (e.g. string vs datetime), making some queries impossible for a few users. --> Karl on support & ill, now making progress, will partially carry over into "slick filters"

New goals

• Data exploration: tiles on dashboards @pauldambra
  Why? Last big piece in data exploration to tie the unified vision together.

• Data exploration: slick filters @mariusandra
  Why? HogQL is powerful and soon in every filter. Make the HogQL editor slick.

• PostHog 3000 prerequisites: continue yeeting ants @Twixes
  Why? Q1 OKR. Also: we ship about 800kb of CSS. About 500kb of that is antd. Not slick.

Sidequests:

• Systematically audit queries @yakkomajuri
• Pivot tables (possibly cont from this week) @mariusandra
• Expand insight person drilldown views
• Other sources of data in live events view (switch to recordings)

[EDsCODE]

Retro:

• Feature Flag Resiliency @neilkakkar
  • Decide updated
• Feature Flag Payloads @EDsCODE
  • Decide endpoint updated
  • JS Lite implemented
  • Frontend in progress/ready for merge?
• UX @liyiy
  • Tags
  • Rollout default fix: new flags default to 0% rollout instead of 100% automatically #13719 (comment)

Planning:

• Rate Limiting (slack thread)
• All other client and server libraries for resiliency and JSON payload

[ellie]

🚀 Infrastructure Team 🚀

Infra hero: @danielxnj

Retro

Sad that Guido will be leaving us, but happy that we've (potentially) filled the role fast

Planning

@guidoiaquinti: Finish up ARM usage, scale up the usage and make sure nothing breaks.
@ellie: Continuing working on postgres! Didn't quite go to plan last time

[tiina303]

Team pipeline🚰

Retro

Goal was: Offsite why? Make sure we carry momentum from offsite.

Jams - Offsite was a ton of fun and very intense. We got quite a bit done. We even did some fire fighting. We laughed. We cried. We bonded.
Tiina - Offsite was very productive and intense. Happy that we have a plan for PoE
Tomas - Offsite was a great introduction to how Team Plumbers works. Also helped me dig into some complicated topics like PoE.
Xavier - Proof that it’s good to be in the same room from time to time.

Sprint planning

Luigi: @xvello

PoE “Person Overrides” ingestion shipped and queries enabled for team 2
why: Query Performance while maintaining Data Quality

Fallback is to focus on Squash logic if queries get held up

[benjackwhite]

Team SessiOS Recordings

Retro

• @benjackwhite Felt like we are less "shipping" than before - good to keep an eye on it.
• @alexkim205 Agrees with rolling out Inspector panel speed, Performance is a little more careful regarding ingestion.

Hang over items from previous sprint

• Internally test and then finish up Unified Network panel to gradually rollout to customers
  (Will be out by end of sprint)

Planning

High priority

• Rollout Network panel to some beta users @alexkim205
  why? We have been running it successfully internally last sprint and the final touches are there so lets get it out.

• Turn on Recordings by default @alexkim205
  why? Huge number of customers we are missing out on as they don't realise it isn't on by default.

• Mobile Session Recordings published (alpha)
  why? Makes sense to test what we have ASAP with real apps (e.g. PostHoger's apps)

Bugs

Low priority / side quests

• Network release todo

  • Look into TTL on CH table

• Ingestion rate limiting @benjackwhite @kappa90 See issue