What actually authorizes requests in Vault is the token. No matter which auth method you use to log in (OIDC, Kubernetes, AppRole, etc.), the client ultimately holds a token issued by Vault, and that token is the embodiment of its permissions.
This article focuses on stable features and covers token fundamentals, type/attribute differences, renewal and revocation, policy integration, and common design patterns. We also flag the points that show up most often on the Associate-level exam.
In Vault, auth backends (such as userpass, OIDC, Kubernetes, and AppRole) are 'login mechanisms,' and they ultimately issue a token to the client. Policies and a TTL are bound to that token, and every subsequent request is evaluated against it.
In other words, the token carries not only the result of authentication but also access control (what can be done on which path), the expiration, and the blast radius of revocation. The exam likes to test the assumption that 'no matter which method you log in with, the token ultimately represents permissions.'
Flow from authentication to token issuance to request evaluation
When a token is issued, the TTL is set along with whether it is renewable, whether it has a max TTL, and whether it has a use-limit. A normal service token simply consumes its TTL, and you can extend it via the renew API (up to the max TTL).
Parent-child relationships govern how revocation propagates. Revoking a parent token simultaneously revokes the child tokens it created (and their grandchildren, and so on). Orphan tokens created with -orphan have no parent, so they are not affected by a parent's revocation. Periodic tokens require renewal on each period interval and operate without a max TTL constraint, but if you stop renewing, they expire at the period boundary.
Vault tokens are mainly split into service tokens and batch tokens, and you can layer attributes on top, such as orphan (no parent) or Periodic (renewal required). It helps to think of the type as governing features and the attribute as governing behavior.
Batch tokens are slimmed down for high-throughput use cases and come with restrictions: not renewable, no child tokens, no accessor, and so on. Service tokens carry the full operational feature set (renew, revoke, accessor, child creation) and suit typical long-running services.
| Type / Attribute | Main characteristics | Renew / Revoke | Parent-child / Accessor |
|---|---|---|---|
| Service Token | Persistent with a rich feature set | Renewable (within max_ttl); explicit revocation supported | Can create children; has accessor; supports Lookup |
| Batch Token | Lightweight, built for high throughput | Not renewable; expires at TTL (explicit revoke and Lookup are limited) | Cannot create children; no accessor |
| Orphan (attribute) | An independent token with no parent | Follows whichever underlying type is used | Parent revocation does not propagate |
| Periodic (attribute) | Requires renewal on every period | No max TTL; expires at the period boundary if renewal stops | Follows the normal parent-child model |
A token holds one or more policies, and each policy defines capabilities such as read, create, update, delete, list, and sudo on a per-path basis. Vault evaluates whether each request is permitted by matching its target path against those policies.
By default the default policy is attached, but you can exclude it at token creation time. The exam frequently tests two points: 'policies are a set of allow rules, and should be designed for least privilege,' and 'you can self-inspect capabilities.'
For services, the baseline is short-lived tokens combined with automatic renewal. Choosing Periodic frees you from the max TTL constraint and makes monitoring easy via renewal health checks. Build in a pattern that fails safe when renewal fails.
Batch tokens shine in high-throughput, read-only proxies and short-running batch jobs. Treat them as non-renewable, keep the TTL strictly short, and rotate by issuing new tokens as needed.
Here are some typical operations. On the exam, knowing the meaning of each option (-policy, -ttl, -period, -use-limit, -orphan, -type=batch) and how renewal, revocation, Lookup, and Accessor relate to each other will be a reliable source of points.
Remember that batch tokens are not renewable, have no accessor, and have restricted Lookup. Periodic tokens must be renewed on every period; stop renewing and they expire at the period boundary.
Common vault token operations
# Log in (e.g., userpass / OIDC)
$ vault login
# Issue a service token (30-min TTL, up to 100 uses, orphan, with the app policy)
$ vault token create -policy=app -ttl=30m -use-limit=100 -orphan
# Periodic token (24-hour period, renewal required)
$ vault token create -policy=svc -period=24h
# Batch token (not renewable, lightweight)
$ vault token create -type=batch -policy=read-only -ttl=5m
# Check capabilities (inspect your own capabilities on a path)
$ vault token capabilities secret/data/*
# Lookup (for service tokens)
$ vault token lookup s.Ku1...
# Retrieve the accessor (prefer logging the accessor, not the token)
$ vault token lookup -format=json s.Ku1... | jq -r .data.accessor
# Revoke by accessor (no need to handle the token string)
$ vault token revoke -accessor A1bC2...
# Renew (renewable service tokens only)
$ vault token renew s.Ku1...
# Revoking a parent token cascades to its descendants
$ vault token revoke s.Parent...
Associate
問題 1
A high-throughput, read-only proxy fetches secrets from Vault. Renewal is not needed, short-lived expiration is fine, and you want to maximize throughput. Which token choice and design fits this use case best?
正解: A
Batch tokens are the best fit for high throughput, short lifetime, and no-renewal needs. They are non-renewable and lightweight, which favors performance. Periodic is for long-running services, routine use of the root token is discouraged, and relying on max-TTL extensions is risky.
Why can Periodic tokens be renewed beyond the max TTL?
Periodic tokens are designed around regular renewal within the period interval. Instead of a max TTL, the model is 'valid as long as it keeps being renewed.' Stop renewing and the token expires at its TTL.
What is a token accessor used for?
It is an identifier that lets you perform Lookup or Revoke without handling the token itself. Accessors are also recorded in audit logs, so accessor-centric operations are safer. Note that batch tokens do not have accessors.
Is the default policy always attached?
It is attached by default, but you can exclude it at token creation time (for example, the CLI's -no-default-policy flag). Following least-privilege design, drop it when it is not needed.
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...