GHSA-28XM-PRXC-5866
Vulnerability from github – Published: 2026-04-23 21:44 – Updated: 2026-04-23 21:44Summary
OpenTelemetry.Sampler.AWS reads unbounded HTTP response bodies from a configured AWS X-Ray remote sampling endpoint into memory.
OpenTelemetry.Resources.AWS reads unbounded HTTP response bodies from a configured AWS EC2/ECS/EKS remote instance metadata service endpoint into memory.
Both of these would allow an attacker-controlled endpoint or be acting as a Man-in-the-Middle (MitM) to cause excessive memory allocation and possible process termination (via Out of Memory (OOM)).
Details
OpenTelemetry.Sampler.AWS
AWSXRaySamplerClient.DoRequestAsync called HttpClient.SendAsync followed by ReadAsStringAsync(), which materializes the entire HTTP response body into a single in-memory string with no size limit. The sampling endpoint is configurable via AWSXRayRemoteSamplerBuilder.SetEndpoint (default: http://localhost:2000).
An attacker who controls the configured endpoint, or who can intercept traffic to it (MitM), can return an arbitrarily large response body. This causes unbounded heap allocation in the consuming process, leading to high transient memory pressure, garbage-collection stalls, or an OutOfMemoryException that terminates the process.
OpenTelemetry.Resources.AWS
The AWSEC2Detector, AWSECSDetector and AWSEKSDetector classes all make HTTP requests to the relevant AWS metadata service (http://169.254.169.254, ECS_CONTAINER_METADATA_URI/ECS_CONTAINER_METADATA_URI_V4 or https://kubernetes.default.svc respectively) to obtain metadata about the running process and its infrastructure.
An attacker who controls the configured endpoint(s), or who can intercept traffic to them (MiTM), can return an arbitrarily large response body. This causes unbounded heap allocation in the consuming process, leading to high transient memory pressure, garbage-collection stalls, or an OutOfMemoryException that terminates the process.
Impact
Denial of Service (DoS). An attacker can destabilize or crash the application by forcing unbounded memory allocation through the X-Ray sampling and/or EC2/ECS/EKS HTTP response paths.
Mitigating Factors
- The default X-Ray sampling endpoint is
http://localhost:2000, which limits remote exposure in default configurations. - Risk increases materially when operators configure the sampler to point at a remote or untrusted endpoint.
Patches
Fixed in OpenTelemetry.Sampler.AWS version 0.1.0-alpha.8 and OpenTelemetry.Resources.AWS version 1.15.1.
The fixes (#4100, #4122) introduce changes that introduce limits to HttpClient requests so that the response body is streamed rather than buffered entirely in memory.
Workarounds
- Ensure the X-Ray sampling endpoint (
http://localhost:2000by default) is not accessible to untrusted parties. - Use network-level controls (firewall rules, mTLS, service mesh) to prevent Man-in-the-Middle (MitM) attacks on the sampling endpoint and/or EC2/ECS/EKS connection.
- If using a remote endpoint, place it behind a reverse proxy that enforces a response body size limit.
{
"affected": [
{
"package": {
"ecosystem": "NuGet",
"name": "OpenTelemetry.Sampler.AWS"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.1.0-alpha.8"
}
],
"type": "ECOSYSTEM"
}
]
},
{
"package": {
"ecosystem": "NuGet",
"name": "OpenTelemetry.Resources.AWS"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.15.1"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-41173"
],
"database_specific": {
"cwe_ids": [
"CWE-770"
],
"github_reviewed": true,
"github_reviewed_at": "2026-04-23T21:44:31Z",
"nvd_published_at": "2026-04-23T19:17:29Z",
"severity": "MODERATE"
},
"details": "### Summary\n\n`OpenTelemetry.Sampler.AWS` reads unbounded HTTP response bodies from a configured AWS X-Ray remote sampling endpoint into memory.\n\n`OpenTelemetry.Resources.AWS` reads unbounded HTTP response bodies from a configured AWS EC2/ECS/EKS remote instance metadata service endpoint into memory.\n\nBoth of these would allow an attacker-controlled endpoint or be acting as a Man-in-the-Middle (MitM) to cause excessive memory allocation and possible process termination (via Out of Memory (OOM)).\n\n### Details\n\n#### OpenTelemetry.Sampler.AWS\n\n`AWSXRaySamplerClient.DoRequestAsync` called `HttpClient.SendAsync` followed by `ReadAsStringAsync()`, which materializes the entire HTTP response body into a single in-memory string with no size limit. The sampling endpoint is configurable via `AWSXRayRemoteSamplerBuilder.SetEndpoint` (default: `http://localhost:2000`).\n\nAn attacker who controls the configured endpoint, or who can intercept traffic to it (MitM), can return an arbitrarily large response body. This causes unbounded heap allocation in the consuming process, leading to high transient memory pressure, garbage-collection stalls, or an `OutOfMemoryException` that terminates the process.\n\n#### OpenTelemetry.Resources.AWS\n\nThe [`AWSEC2Detector`](https://github.com/open-telemetry/opentelemetry-dotnet-contrib/blob/171c6b81f88831641b56b470e6f92862e605013d/src/OpenTelemetry.Resources.AWS/AWSEC2Detector.cs), [`AWSECSDetector`](https://github.com/open-telemetry/opentelemetry-dotnet-contrib/blob/171c6b81f88831641b56b470e6f92862e605013d/src/OpenTelemetry.Resources.AWS/AWSECSDetector.cs) and [`AWSEKSDetector`](https://github.com/open-telemetry/opentelemetry-dotnet-contrib/blob/171c6b81f88831641b56b470e6f92862e605013d/src/OpenTelemetry.Resources.AWS/AWSEKSDetector.cs) classes all make HTTP requests to the relevant AWS metadata service (`http://169.254.169.254`, `ECS_CONTAINER_METADATA_URI`/`ECS_CONTAINER_METADATA_URI_V4` or `https://kubernetes.default.svc` respectively) to obtain metadata about the running process and its infrastructure.\n\nAn attacker who controls the configured endpoint(s), or who can intercept traffic to them (MiTM), can return an arbitrarily large response body. This causes unbounded heap allocation in the consuming process, leading to high transient memory pressure, garbage-collection stalls, or an `OutOfMemoryException` that terminates the process.\n\n### Impact\n\nDenial of Service (DoS). An attacker can destabilize or crash the application by forcing unbounded memory allocation through the X-Ray sampling and/or EC2/ECS/EKS HTTP response paths.\n\n### Mitigating Factors\n\n- The default X-Ray sampling endpoint is `http://localhost:2000`, which limits remote exposure in default configurations.\n- Risk increases materially when operators configure the sampler to point at a remote or untrusted endpoint.\n\n### Patches\n\nFixed in `OpenTelemetry.Sampler.AWS` version `0.1.0-alpha.8` and `OpenTelemetry.Resources.AWS` version `1.15.1`.\n\nThe fixes (#4100, #4122) introduce changes that introduce limits to `HttpClient` requests so that the response body is streamed rather than buffered entirely in memory.\n\n### Workarounds\n\n- Ensure the X-Ray sampling endpoint (`http://localhost:2000` by default) is not accessible to untrusted parties.\n- Use network-level controls (firewall rules, mTLS, service mesh) to prevent Man-in-the-Middle (MitM) attacks on the sampling endpoint and/or EC2/ECS/EKS connection.\n- If using a remote endpoint, place it behind a reverse proxy that enforces a response body size limit.",
"id": "GHSA-28xm-prxc-5866",
"modified": "2026-04-23T21:44:31Z",
"published": "2026-04-23T21:44:31Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/open-telemetry/opentelemetry-dotnet-contrib/security/advisories/GHSA-28xm-prxc-5866"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-41173"
},
{
"type": "WEB",
"url": "https://github.com/open-telemetry/opentelemetry-dotnet-contrib/pull/4100"
},
{
"type": "WEB",
"url": "https://github.com/open-telemetry/opentelemetry-dotnet-contrib/pull/4122"
},
{
"type": "PACKAGE",
"url": "https://github.com/open-telemetry/opentelemetry-dotnet-contrib"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:H",
"type": "CVSS_V3"
}
],
"summary": "OpenTelemetry.Sampler.AWS \u0026 OpenTelemetry.Resources.AWS have unbounded HTTP response body reads"
}
Sightings
| Author | Source | Type | Date |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.