There was an error while loading. Please reload this page.
docs: rewrite networking/isolation page for public audience Prior draft was written at internal-engineering detail level (file:line citations, internal repo names, exact env var/annotation names, an attacker-reasoning walkthrough) - appropriate for verification rigor, wrong register for a public customer-facing page. Rewritten to state the same underlying facts (no per-tenant network segmentation, no per-tenant CPU/memory limits, 64MB request size, 60s timeout, no bandwidth quota by default) in plain customer terms, matching the AWS/GCP shared-responsibility doc style. The fully-cited technical version is preserved internally at agent-team-osc-dev/docs/platform-networking-isolation-model.md rather than discarded - it's the source of truth backing this page and the team's re-verification reference. birme feedback: too detailed a level for public documentation. Co-Authored-By: Claude <noreply@anthropic.com>
docs: add OSC platform networking and tenant isolation model page Sourced from infra-osaas @275182d, osaas-deploy-manager @c06f6af, osaas-lib-orchestrator @32b9811, plus live read-only kubectl verification against both production clusters (Linode LKE and Elastx prod-se). Addresses Eyevinn/osaas-ai#1543. Security-reviewed before publish: 2 factual errors found and fixed (proxy-body-size misattributed from the unrelated image-server ingress; ingress-osaas-myapps-nginx name corrected to the actual IngressClass osaas-myapps-nginx), both TBD callouts resolved with live cluster evidence (no CNI-level tenant segmentation confirmed on either cluster; tenant pod RBAC confirmed to have no cluster enumeration path), and one confidence-calibration edit on tenant-ID guessability. Co-Authored-By: Claude <noreply@anthropic.com>