TLDR DevOps 2026-10-23
ObservabilityCON 🔭, Harness AI Budgets 💰, Scaling GitOps 🧗
Introducing Harness AI Budgets (5 minute read)
Harness AI Budgets turns AI cost policies into enforced weekly spending limits scoped to a team, developer, model, or project. Access pauses at 100% until the weekly reset or an approved increase. In Harness' internal test, weekly AI spend per developer fell about 37%.
Critical PHP RCE Vulnerabilities: Patch Now (CVE-2026-109618, CVE-2026-109619) (5 minute read)
Two critical remote code execution vulnerabilities in PHP, tracked as CVE-2026-109618 and CVE-2026-109619, were publicly disclosed on October 21. Proof-of-concept exploit code is available, and the flaws affect every PHP version from PHP 5 onward. DigitalOcean is automatically patching WordPress instances on its Cloudways managed product, but customers running PHP on Droplets or other unmanaged services must update to PHP 8.2.35, 8.3.36, 8.4.27, or 8.5.12 immediately. There are no patches for PHP 8.1 or earlier.
ObservabilityCON 2026: A guide to all the announcements from Grafana Labs (10 minute read)
Grafana Labs announced Grafana Cloud updates covering AI operations, cost controls, and full-stack context. Assistant Watchers reached general availability, Adaptive Telemetry customers report cutting spend 30% to 50% on average, and the Grafana Marketplace launched with more than 30 plugins.
Scaling GitOps With Release Promotion (4 minute read)
Jesse Suen of Akuity describes a fleet of over 200 clusters where every merge synced everywhere with no gate. He argues sync tools do not handle release tracking or promotion decisions and recommends staged promotion that rolls out in waves and halts on failed checks.
I Don't Trust ON DELETE CASCADE (6 minute read)
Large cascading deletes can turn a routine schema migration into an outage by holding locks while new queries queue behind the waiting migration. Soft deletes followed by bounded background cleanup keep transactions smaller, with additional pitfalls around unindexed foreign keys and logical replication.
Presentation: AI is an Amplifier: Scale High Performance, Not Dysfunction (33 minute read)
Nathen Harvey of Google Cloud's DORA team presents 2025 survey findings from nearly 5,000 respondents, about 90% of whom use AI professionally. He argues AI amplifies existing organizational strengths and dysfunctions, and introduces the DORA AI Capabilities Model.
8 eMudhra certificates are compromised. Here's what you need to know (4 minute read)
Private keys for eight widely trusted eMudhra intermediate certificates were compromised on October 19, allowing attackers to forge certificates for any domain. Cloudflare has patched its CDN, Zero Trust, and Tunnel products. Customers must update the Cloudflare One Client and 1.1.1.1 app to specific versions to remain protected.
Before you evaluate an AI SRE, give your existing agent context (6 minute read)
Engineers at Radar tested whether adding context to coding agents improves Kubernetes diagnostics. When running 50 fault scenarios, agents with context reached a correct diagnosis within two minutes on 86% of faults, compared to 60% for agents using a standard shell. Accuracy improved only slightly with context, but the time savings were significant. The test used Claude Code and Microsoft's SREGym benchmark on a live Amazon EKS cluster.
Get our free daily newsletter with curated tools 💻, trends 📈, and insights 💡, for DevOps Engineers 👨💻
Join 350,000 readers for
one daily email