How long are tokens valid and why the split?
The API and mobile apps use JWT tokens, and the browser uses a session. The access token is valid for 12 hours. The refresh token is valid for 7 days and is single-use: after use it is exchanged for a new pair. A short access token limits the effects of a leak, while a longer refresh token means the user does not have to log in every day.
For a person with a phone this is a convenient solution, because when refresh fails the system shows the login screen. For a wearable device the same mechanism can be a trap, described in the next section.
What happens when a watch cannot log in?
A kiosk device, such as a watch without a keyboard, cannot log in again after its token expires. It cannot type a password or code, so it stops sending signals. The dispatcher sees the employee's missing refresh on the map, and the cause is the authentication state, not the battery and not the network.
Because the refresh token is single-use, every failed refresh, for example after a longer connectivity gap, can close the way back. That is why devices of this type need separate, explicit treatment in the configuration.
Validity of tokens and trusted device memory
What is the kiosk device account flag?
It is a separate account flag that is switched on explicitly and indicates that the account belongs to a kiosk device. An account with the flag uses the token grace period, a short window after expiry during which the device can still renew access without a person's involvement.
The flag covers only accounts indicated by the administrator. Other accounts keep standard expiry and the requirement to log in again, so the overall security level of the installation does not change.
After a system update, do device accounts get the flag automatically?
No. The migration that adds the flag deliberately does not assign it to any account. Account identifiers differ between installations, and granting permissions during a migration would be a silent change that the plant has not approved.
In practice the plant administrator, usually with the IT department, indicates the device fleet accounts and switches the flag on deliberately. This is a good moment to write down a list of devices, their owners and where they are kept.
How to check which devices use the grace period?
Every admission and every refusal after the device token grace period is exceeded is written to the log with a warning level, the account and the expiry date. Comparing these entries with the list of fleet accounts shows which devices work at the edge of token validity.
For IT this is also a tool for counting fleet usage: a device that appears in the log every day needs a connectivity check or a battery replacement before it becomes a failure visible to the dispatcher.
Do MFA and trusted devices apply to watches?
MFA is an installation option with an email code valid for 10 minutes and TOTP, and a trusted device remembers the verification for 7 days. This concerns people logging in through the interface. A kiosk device without a keyboard does not go through this process, so its access relies on an account with the flag, not on a code.
The rule for IT: people's accounts are protected by MFA, and device accounts are protected by limiting their permissions, the flag granted only to indicated accounts, and log supervision.
What should the plant procedure record?
The procedure should name the person responsible for the list of device accounts, the review cycle for the warning log, and the course of action when a device goes silent: check the battery and the network first, and only then the account. Recording this order shortens diagnosis time and reduces the number of tickets sent to the IT department.
Video
Frequently asked questions
Does the kiosk account flag lower the security of the whole installation?
No, because it covers only accounts indicated by the administrator. Other accounts keep standard token expiry and the requirement to log in again.
Can a refresh token be used several times?
No. It is single-use: after use it is exchanged for a new pair of tokens. This limits the effects of intercepting an old token.
Where can refusals after the grace period be seen?
In the log. The entry has a warning level and includes the account and the expiry date, which makes it possible to count fleet usage.
Sources
Let's set up login for the device fleet
We will go through device accounts that have no keyboard with your IT team, agree which accounts get the flag, and decide who reads the log.
Let's talk