drills/ — DR drill reports¶
Per carrier-noc-runbook.md §2, restore drills run quarterly for every component with a published RTO/RPO target. One Markdown file per drill, named drill-YYYY-MM-DD-<scenario-slug>.md.
Required scenarios (quarterly cycle)¶
| Quarter offset | Scenario | Slug |
|---|---|---|
| Q1 | RDS PITR restore to T−1 h on a fresh instance | rds-pitr-restore |
| Q2 | Cross-region failover dry-run (sa-east-1 → us-east-1 ALB + RDS read replica promotion) | region-failover |
| Q3 | S3 cross-region replication validation (delete-and-restore on a sentinel object) | s3-replication |
| Q4 | Audit-log restore from KMS-encrypted backup + integrity check | audit-log-restore |
Drills can be re-ordered between quarters as long as each scenario runs once per 12 months.
Required fields in every drill report¶
- Scenario, slug, start/end timestamps in
Z(UTC). - RTO actual vs target; RPO actual vs target.
- Operator names (lead + observer).
- Deviations from procedure with rationale.
- Follow-up actions filed as GitHub issues (link each).
- Verdict:
pass/fail/pass-with-followups.
Index¶
Maintained manually. Newest at top.
| Date | Slug | Verdict | Notes |
|---|---|---|---|
| (no drills yet — first drill due before first Carrier contract effective date) |
Gitignore¶
Like ../carrier/, this directory does not commit individual drill reports to the public repo (they reference internal tooling URLs, vendor incident IDs, sometimes customer slugs). Only README.md and .gitignore are tracked. Account owner is responsible for backing up the reports to the internal shared drive.