EBS snapshot audit and lifecycle policy

[dashinfraclaw]

Problem

Both AWS accounts accumulated large numbers of old EBS snapshots with no lifecycle policy. dashevo had 239 snapshots (17TB) costing $240-400/mo — all deleted on 2026-03-20. dashdev still has ~129 snapshots that need review.

Completed

• dashevo: 239 snapshots deleted (old blockchain bcsync data from 2021-2023)
• dashevo: 11 custom AMIs deregistered + backing snapshots deleted

Remaining

• dashdev: ~129 snapshots — audit needed
  • Which are backing AMIs still in use?
  • Which are old/orphaned blockchain data?
  • Estimated cost: ~$210/mo (from billing analysis)

Prevention

Implement EBS snapshot lifecycle policies:

1. AWS Data Lifecycle Manager (DLM): Create policies to auto-delete snapshots older than X days
2. Tag-based retention: All snapshots must have retention-days tag
3. Monthly audit: Cron job to report untagged/old snapshots
4. Terraform: Manage DLM policies in dash-infra (Create dash-infra Terraform project for shared services #21)

Related

• Consolidate dashevo AWS account into dashdev #19 (dashevo consolidation — snapshots already cleaned)
• Create dash-infra Terraform project for shared services #21 (dash-infra should manage lifecycle policies)

[dashinfraclaw]

dashdev Snapshot Cleanup — 2026-03-30

Completed

• Audited all 54 EBS snapshots in dashdev (us-west-2)
• Deleted 38 orphaned snapshots (3,830 GB total):
  • 1 large volume resize artifact from 2021 (2 TB)
  • 3 orphaned snapshots from previously deleted AMIs
  • 1 stale DashCore testnet datadir (v0.17.0.3, June 2022)
  • 1 debug snapshot
  • 32 HP masternode snapshots from June 2025 batch operation
• Verified 16 remaining snapshots are AMI-backed

Still TODO

• 16 unused AMIs identified (gitian builders, old network templates, benchmarks — all 2021-2022 era, none backing running instances). Need ec2:DeregisterImage permission on infraclaw-role to clean up
• Implement DLM lifecycle policies to prevent re-accumulation
• HP masternode backup strategy (see new issue)

Prevention

Once the AMI cleanup is done, the remaining work is setting up automated lifecycle policies per the original issue description.