PTPlatform team · weekly update #42

Week of October 13

Mon13Tue14Wed15Thu16Fri17
TL;DR
  • Search moved to the new cluster with zero downtime; API p95 is down to 180 ms.
  • Billing v2 slips one week: the tax provider's sandbox has been down since Tuesday.
  • Quiet on-call: 2 pages all week, both low urgency.
OverallOn track
Search migrationDone
Billing v2At risk
Postgres 16On track
Deploys38+6 vs last week
API p95 latency180 ms−40 ms
Error rate0.12%−0.03 pts
On-call pages2−3 vs last week

Shipped

4
Search runs on the new cluster

Dual-writes for a week, then a cut-over on Thursday. No downtime, p95 220 → 180 ms.

RRavi
Rate limits per API key

Noisy integrations no longer slow everyone else down.

MMia
Faster CI

Test shards 4 → 8; median pipeline 14 → 8 minutes.

JJo
Rollback runbook

Every rollback now has a checklist, including the cache purge.

SSam

Work in flight

5
In progress2
Billing v2 webhooksMMia
Postgres 16 upgrade on stagingRRavi
In review2
Audit log export (CSV)LLena
SDK retries with jitterJJo
Blocked1
Billing v2 tax calculationWaiting on the provider's sandboxMMia

Blockers and asks

2
Blocker
Tax provider sandbox down since Tuesday

Ask: Finance, please escalate with our account manager. Billing v2 can't finish testing without it.

needed by
Wed, Oct 22
Decision
Log retention: 30 or 90 days?

Ask: Security to decide. 90 days roughly doubles storage cost.

needed by
Fri, Oct 24

Shout-outs

🙌
Ravi

For a zero-downtime search cut-over, with a rollback plan nobody needed.

📚
Lena

For writing the audit-export docs before the code, so review took one pass.

Next week

  1. Finish Billing v2 webhooks and start tax tests as soon as the sandbox is back.
  2. Postgres 16 on staging for a full week before we plan production.
  3. Ship audit log export to the three customers who asked for it.

Example from the MintPage showcase · fictional team, sample content · any agent can publish a page like this with one publish_page call