Vault always starts in a "Sealed" state right after the process boots. Only after an unseal opens the Barrier can it serve any secrets.
This article walks through the boot → unseal → HA activation flow and contrasts the Shamir (manual) approach with the KMS/HSM (Auto Unseal) approach, from both an operations and an exam-prep perspective.
Vault reads its config file and initializes storage (integrated Raft storage, Consul, etc.), the listener, and the "seal" stanza (Shamir by default). The process always boots Sealed, and the encryption Barrier must be opened via unseal before Vault can serve traffic.
Once unseal completes, HA setups run a leader election: one node becomes Active and the rest are Standby. With Auto Unseal, each node queries the KMS/HSM and opens the Barrier automatically — no manual intervention required.
Vault startup → unseal → HA flow (conceptual diagram)
[Client]
|
[Vault Process] --read--> [Config: storage/listener/seal]
|
v
[Sealed Barrier]
/ \
/ \
manual auto
(SHAMIR) (KMS/HSM)
| |
v v
[Unseal Shares t/n] [KMS decrypt barrier key]
\ /
v v
[Barrier Opened]
|
v
[HA Election]
/ \
v v
[Active] [Standby]
| |
serve writes proxy/reads* (*構成による)Vault encrypts data in storage with the "Barrier" and protects the key material that opens it. The master key generated at initialization (operator init) is split using Shamir's Secret Sharing (in the manual case). Once the threshold number of shares is supplied, the master key is reconstructed and the Barrier opens.
With Auto Unseal, an external key manager (KMS/HSM) protects and decrypts the master key, so the Barrier opens automatically at startup. In this mode Vault issues "Recovery Keys" instead, which are used for specific operations such as disaster recovery or generating a new root token.
Manual unseal requires entering threshold-many of the unseal keys produced at initialization to open the Barrier. In an HA cluster, every node must be unsealed individually (both Active and Standby).
Key management is the cornerstone of operations. Design the share count and threshold around your team size and BCP requirements, and plan for key storage, distribution, and rotation (rekey).
A typical manual unseal sequence (example)
# 初期化(共有数5、しきい値3)※クラスタで一度のみ
vault operator init -key-shares=5 -key-threshold=3
# => Unseal Key 1..5 と 初期rootトークンが出力される(安全に保管)
# Unseal(1台につき3回、異なるキーを入力)
vault operator unseal
vault operator unseal
vault operator unseal
# 状態確認
vault status
# Sealed: false になっていればUnseal完了
# しきい値の変更や共有の再生成(ローテーション)
vault operator rekey -key-shares=5 -key-threshold=3
# 手動で再Sealする場合(緊急時の遮断など)
vault operator sealAuto Unseal protects Vault's master key with an external KMS/HSM, so each node opens the Barrier automatically at startup. It reduces manual operations and the burden of restarts and scale-out events.
Under this model, Vault issues Recovery Keys instead of unseal keys. They are required for specific maintenance operations such as generate-root and recovery rekey, so design a clear distributed-storage policy for them.
Auto Unseal configuration snippet (AWS KMS example)
# /etc/vault.d/config.hcl(抜粋)
storage "raft" {
path = "/opt/vault/data"
node_id = "vault-1"
}
listener "tcp" {
address = "0.0.0.0:8200"
tls_disable = 0
tls_cert_file = "/etc/vault.d/certs/server.crt"
tls_key_file = "/etc/vault.d/certs/server.key"
}
seal "awskms" {
region = "ap-northeast-1"
kms_key_id = "arn:aws:kms:ap-northeast-1:123456789012:key/xxxxxxxx-xxxx-xxxx-xxxx-xxxxxxxxxxxx"
}
api_addr = "https://vault-1.example.com:8200"
cluster_addr = "https://vault-1.example.com:8201"In an HA setup (Raft/Consul), every node is expected to be unsealed. The Active node leads writes, and Standbys proxy requests or serve reads (depending on configuration and features). Nodes that remain sealed cannot join the cluster and undermine failover resilience.
With Auto Unseal, each node queries the KMS/HSM independently and unseals itself. With manual unseal, every node needs the threshold number of key entries. For rolling restarts, either automating unseal (Auto Unseal) or maintaining a solid runbook pays off.
Example: checking HA status
# HA/Barrierの状態確認
vault status
# 出力例
# Sealed: false
# HA Enabled: true
# HA Cluster: https://vault-1.example.com:8201
# HA Mode: active (他ノードは standby)For Associate and Ops exams, common topics include the differences between manual and Auto Unseal, the t/n threshold, the per-node unseal requirement in HA, and behavior during a KMS outage. Lock in the basics of operator init / unseal / rekey / seal / status, and make sure you can distinguish the role of "Recovery Keys" under Auto Unseal (used for generate-root and recovery rekey).
In production, clearly define the threshold based on BCP requirements, document key custody and procedures, and monitor KMS reachability (VPC endpoints, firewalls).
| Item | Manual Unseal (Shamir) | Auto Unseal (KMS/HSM) | Notes |
|---|---|---|---|
| Startup operation | Key holders enter t shares | Each node decrypts automatically via KMS/HSM | Auto Unseal is more operationally efficient |
| What gets distributed | Unseal keys (n shares, threshold t) | Recovery Keys (held in distributed custody) | Note that their purposes differ |
| External dependencies | None (human-only) | KMS/HSM plus network and IAM | Availability design matters |
| HA cluster | Each node requires t key entries | Each node unseals automatically | Rolling restarts are noticeably easier with Auto Unseal |
| Failure-mode risk | Losing keys means unseal is impossible | If the KMS is unreachable, newly started nodes cannot unseal | Already-unsealed nodes keep running |
| Rotation | vault operator rekey | vault operator rekey -recovery, etc. | Always sync procedure changes with custody updates |
Associate / Ops
問題 1
An organization runs a 3-node Vault cluster (Raft) with Auto Unseal (AWS KMS) enabled. During a rolling restart, two nodes successfully unsealed but the third cannot reach the KMS and remains Sealed. Which statement about this situation is correct?
正解: A
With Auto Unseal, each node opens the Barrier independently via the KMS. Nodes that can reach the KMS unseal and can form an Active/Standby pair in HA. Once connectivity is restored, the remaining node unseals automatically. Recovery Keys are for maintenance operations (generate-root, recovery rekey, etc.) — they do not substitute for the KMS during a regular unseal.
Where and how many times should I run vault operator init?
Exactly once per cluster, on the first node. Subsequent nodes join the cluster and each performs its own unseal. Never initialize twice — doing so corrupts the Barrier and destroys data.
Can I migrate from manual unseal to Auto Unseal?
Yes. Add a new seal stanza (for example awskms) and rekey to rewrap the master key. Perform the migration while Vault is unsealed, and prepare a runbook, backups, and a rollback plan.
What should I check with vault status?
Check Sealed (true/false), Initialized, HA Enabled, and HA Mode (active/standby). Sealed: true is normal right after startup and flips to false once unseal completes. In HA setups, also track the Active/Standby role as part of your monitoring.
Practice with certification-focused question sets
無料で問題を解いてみるNicheeLab Editorial Team
NicheeLab editorial team focused on data engineering and cloud certification learning. Content is structured around practical study needs and official exam domains.
Vault Core Concepts: Sealed/Unsealed, Auth, Secrets (2026)
Vault fundamentals — sealed/unsealed state, auth methods, se...
Vault Operations Professional (VOP-003): Complete Guide (2026)
Pass the Vault Operations Professional exam — enterprise pat...
Vault Path-Based Routing: API URL Structure (2026)
How Vault's path-based routing works — mount points, sub-pat...
Vault Tokens: Auth Token Mechanics (2026)
Token fundamentals — service vs. batch tokens, accessor, ren...
Vault Token Types: Service, Batch, Periodic (2026)
Service vs. batch tokens compared — performance, ACL behavio...