25 writeups — drag to explore
01 · build2026
One INSERT, 2.8M rows, no ETL
02 · incident2026
The 17th rule never shipped
03 · incidentSEV1
Every certificate expired at midnight
04 · build2026
One NAT instance per AZ, not one per VPC
05 · plan2026
One t3.nano between us and the internet
06 · incidentSEV1
The night DNS lied to us
07 · note2026
Autoscaling can't catch a burst
08 · note2026
depends_on on a module replaced my NAT
09 · note2026
dnf installed a different Postgres
10 · note2026
It looked like a rate limit
11 · note2026
Traefik ignores your certificateRefs
12 · incident2026
110 pods, zero reservations
13 · build2026
Postgres 12 to 16 with no downtime
14 · postmortemSEV2
One bad limit and the whole node went with it
15 · build2026
CI runners that scale to zero
16 · incidentSEV1
The autoscaler scaled us down at peak
17 · build2025
Taking 41% off the AWS bill
18 · build2025
Cutting the build from 34 minutes to 6
19 · postmortemSEV1
The etcd volume that quietly filled up
20 · build2025
Moving three clusters to GitOps
21 · incidentSEV2
Out of ports on the NAT gateway
22 · note2025
Page on symptoms, not causes
23 · postmortemSEV2
The consumer lag that fed itself
24 · postmortemSEV2
A Terraform plan that looked like a no-op
25 · build2024