44 lines
2.7 KiB
Markdown
44 lines
2.7 KiB
Markdown
# DR Plan v3 — Response to Third-Party Reviewers
|
|
|
|
## Two reviewers, two passes. All findings addressed.
|
|
|
|
### Reviewer 1 (Technical DR Specialist)
|
|
|
|
| Finding | Status | Addressed In |
|
|
|---|---|---|
|
|
| Failover timing mismatch (5min check can't detect 2min outage) | Fixed | Corrected to 30s checks, 4 failures to trigger = 2min detection, ~3-3.5min actual RTO, 5min committed target |
|
|
| Backup standard says daily but not happening | Flagged | Section 3.1 backup matrix shows live vs gap status; P0 items for Docker volume + DB dump automation |
|
|
| AI server 92% disk critical | Flagged | P0 action item #1 |
|
|
| Memory consolidation data loss risk | Mitigated | S3 backup-before-prune + PROTECT regex expansion + restore procedure documented |
|
|
| Tony's Hermes undocumented | Flagged | P1 action item #7 |
|
|
|
|
### Reviewer 2 (Systems Management SME)
|
|
|
|
| Finding | Status | Addressed In |
|
|
|---|---|---|
|
|
| S3 single point of failure | Mitigated | Two-destination design (Wasabi us-east-1 primary + Wasabi us-west-2 or Backblaze B2 secondary) |
|
|
| Backup frequency vs RPO mismatch | Fixed | Per-data-type RPO: 15min (Hermes state), 24h (Docker/data), 24h backups |
|
|
| Restore procedures not detailed | Fixed | Section 6 — per-server runbooks with exact commands |
|
|
| Failback under-defined | Fixed | Section 4.4 — 12-step protocol with rollback, no auto-failback |
|
|
| No backup restore testing | Fixed | Section 7 — testing schedule with cadence, log template, sign-off table |
|
|
| Retention too short (48h snapshots) | Flagged | P1 item #9 — extend to 30 days |
|
|
| Backup security not described | Fixed | Section 3.2 — Object Lock, write-only IAM, credential separation |
|
|
| Application backup details vague | Partial | Section 2.3 — data paths defined per app |
|
|
| Secrets recovery not documented | Fixed | Section 1 — shared dependencies table with credential locations |
|
|
| Provider/account outage missing | Fixed | Section 4.5 — netcup account failure scenario |
|
|
|
|
## Still Pending (Not Yet Implemented, Only Designed)
|
|
|
|
| Item | Status | Required For Sign-Off |
|
|
|---|---|---|
|
|
| Object Lock enabled on buckets | Designed, not done | Yes |
|
|
| Write-only IAM credentials created | Designed, not done | Yes |
|
|
| S3 secondary backup destination provisioned | Designed, not done | Yes |
|
|
| Docker volume backup automation deployed | Flagged, not done | Yes |
|
|
| MariaDB dump cron deployed (wphost02, fleettracker360) | Flagged, not done | Yes |
|
|
| 30-second watchdog deployed on app1-bu | Designed, not done | Yes |
|
|
| Tony's Hermes documented + backed up | Flagged, not done | No (non-critical) |
|
|
| Quarterly failover/failback test performed | Scheduled, not done | Yes |
|
|
| One database restore test performed | Scheduled, not done | Yes |
|
|
| One Docker volume restore test performed | Scheduled, not done | Yes |
|