NATS nats-server before 2.9.23 and 2.10.x before 2.10.2 has an authentication bypass. An implicit $G user in an authorization block can sometimes be used for unauthenticated access, even when the intention of the configuration was for each user to have an account. The earliest affected version is 2.2.0.
NATS nats-server before 2.9.23 and 2.10.x before 2.10.2 has an authentication bypass. An implicit $G user in an authorization block can sometimes be used for unauthenticated access, even when the intention of the configuration was for each user to have an account. The earliest affected version is 2.2.0.
NATS nats-server before 2.9.23 and 2.10.x before 2.10.2 has an authentication bypass. An implicit $G user in an authorization block can sometimes be used for unauthenticated access, even when the intention of the configuration was for each user to have an account. The earliest affected version is 2.2.0.
## Background
NATS.io is a high performance open source pub-sub distributed communication technology, built for the cloud, on-premise, IoT, and edge computing.
NATS users exist within accounts, and once using accounts, the old authorization block is not applicable.
## Problem Description
Without any authorization rules in the nats-server, users can connect without authentication.
Before nats-server 2.2.0, all authentication and authorization rules for a nats-server lived in an "authorization" block, defining users. With nats-server 2.2.0 all users live inside accounts. When using the authorization block, whose syntax predates this, those users will be placed into the implicit global account, "$G". Users inside accounts go into the newer "accounts" block.
If an "accounts" block is defined, in simple deployment scenarios this is often used only to enable client access to the system account. When the only account added is the system account "$SYS", the nats-server would create an implicit user in "$G" and set it as the `no_auth_user` account, enabling the same "without authentication" logic as without any rules.
This preserved the ability to connect simply, and then add one authenticated login for system access.
But with an "authorization" block, this is wrong. Users exist in the global account, with login rules. And in simple testing, they might still connect fine without administrators seeing that authentication has been disabled.
The blind-spot on our part came from encouraging and documenting a switch to using only "accounts", instead of "authorization".
In the fixed versions, using an "authorization" block will inhibit the implicit creation of a "$G" user and setting it as the `no_auth_user` target. In unfixed versions, just creating a second account, with no users, will also inhibit this behavior.
## Affected versions
NATS Server:
* 2.2.0 up to and including 2.9.22 and 2.10.1
* Fixed with nats-io/nats-server: 2.10.2 and backported to 2.9.23
## Workarounds
In the "accounts" block, define a second non-system account, leave it empty.
accounts {
SYS: {
users: [
{ user: sysuser, password: makemeasandwich }
]
}
DUMMY: {} # for security, before 2.10.2
}
system_account: SYS
## Solution
Any one of these:
1. Upgrade the NATS server to at least 2.10.2 (or 2.9.23)
2. Or define a dummy account
3. Or complete the migration of authorization entries to be inside a named account in the "accounts" block
## Credits
Problem reported by Alex Herrington.
Addressed publicly in a GitHub Discussion prior to this advisory.
Without any authorization rules in the nats-server, users can connect without authentication.
Before nats-server 2.2.0, all authentication and authorization rules for a nats-server lived in an "authorization" block, defining users. With nats-server 2.2.0 all users live inside accounts. When using the authorization block, whose syntax predates this, those users will be placed into the implicit global account, "$G". Users inside accounts go into the newer "accounts" block.
If an "accounts" block is defined, in simple deployment scenarios this is often used only to enable client access to the system account. When the only account added is the system account "$SYS", the nats-server would create an implicit user in "$G" and set it as the "no_auth_user" account, enabling the same "without authentication" logic as without any rules.
This preserved the ability to connect simply, and then add one authenticated login for system access.
But with an "authorization" block, this is wrong. Users exist in the global account, with login rules. And in simple testing, they might still connect fine without administrators seeing that authentication has been disabled.
In the fixed versions, using an "authorization" block will inhibit the implicit creation of a "$G" user and setting it as the "no_auth_user" target. In unfixed versions, just creating a second account, with no users, will also inhibit this behavior.
EUVD source date text has no stated timezone and is retained without inventing a UTC instant. Related bulletin membership is not an alias, common cause or applicability claim. A missing vector with a zero score is treated as unrated. Complete vendor/product and underlying source evidence remains in the original JSON.
Original publisher date fields
datePublished: Oct 30, 2023, 12:00:00 AM
dateUpdated: Sep 9, 2024, 8:30:00 PM
1 related source bulletins · separate from aliases
msrc_CVE-2023-47090: NATS nats-server before 2.9.23 and 2.10.x before 2.10.2 has an authentication bypass. An implicit $G user in an authorization block can sometimes be used for unauthenticated access even when the intention of the configuration was for each user to have an account. The earliest affected version is 2.2.0.
CVE-2023-47090
No normalized affected-package data is projected for this source revision. The original publisher record may contain additional product and version evidence; this does not establish that a package is unaffected.
NVD analysis and CPE criteria are source context, not proof that a package or your environment is affected. Complete affected data, logical configurations and SSVC are retained in the original JSON.
The original JSON retains the logical configuration operators, exclusions and version bounds.
No normalized affected-package data is projected for this source revision. The original publisher record may contain additional product and version evidence; this does not establish that a package is unaffected.