CVE-2026-111115
Low risk· EPSS 15%Exploitation Probability (EPSS)
Low risk15th percentile - higher than 15% of all known CVEs
Summary
A vulnerability in Apache Kafka's code-generated readers for flexible protocol versions reads a tagged-field count declared by the sender and retains a RawTaggedField object for each unknown field. The count was bounded only by the outer request frame size, not by remaining buffer bytes or a fixed cap, allowing a modest request to declare tens of millions of tagged fields and force one object allocation per field. Flexible request headers are parsed before authorization, so the amplification applies pre-authentication.
Risk Assessment
An attacker can remotely cause a denial of service on the Kafka broker by sending a request with a huge number of tagged fields in the header, leading to heap exhaustion or GC stalls and cluster unavailability. The risk is high because no authentication or ACLs are required.
Recommendation
Upgrade to Apache Kafka 4.2.2, 4.3.2, or 4.4.0. Until then, lowering socket.request.max.bytes and restricting network access to broker listeners reduces, but does not remove, exposure.
Other vulnerabilities in Apache Kafka
See all- CVE-2026-33557Critical
In Apache Kafka, the default JWT validator class (`DefaultJwtValidator`) accepts any JWT token without validating its signature, issuer, or audience. An attacker can generate a token from any issuer with any username, and the broker will accept it.
- CVE-2026-83525Low risk· EPSS 8%
Incorrect operator (CWE-480) in the FetchSnapshot RPC handler in Apache Kafka from 2.8.0 before 4.2.2, and from 4.3.0 before 4.3.2, allows an authenticated attacker with CLUSTER_ACTION privilege to cause a denial of service. Exploitation occurs via a crafted FetchSnapshotRequest with an empty topics array or a topic with no partitions, terminating the controller process.
- CVE-2026-114349Low risk· EPSS 16%
A vulnerability in Apache Kafka allows an authenticated producer with Write permission on a topic to declare a decompressed body size of up to about 2 GiB in a compressed record, forcing a buffer allocation and causing an OutOfMemoryError. This can crash broker request-handler threads, consumers, and permanently stop the log-cleaner thread on compacted topics.
- CVE-2026-112375Low risk· EPSS 13%
An authenticated Kafka principal with no authorization grants can make a broker read a file or environment variable from the broker host and return its contents. An IncrementalAlterConfigs or AlterConfigs request targeting a BROKER resource reaches configuration validation before any authorization check, and configuration-variable resolution (${file:...}, ${env:...}) bypasses the config-providers allowlist added as mitigation for CVE-2024-31141.
- CVE-2026-111551Low risk· EPSS 13%
Several Apache Kafka deserialization paths read a size or count field from input data and use it directly to allocate an array, collection, or byte buffer without checking the bytes actually available. A small crafted input can declare a very large size and force a multi-gigabyte allocation, causing an OutOfMemoryError. Affected paths include ProcessorRecordContext, ChangedDeserializer, CombinedKeySchema, ProcessorMetadata, and AssignmentInfo in Kafka Streams, and ListDeserializer in the clients library.
- CVE-2026-111108Low risk· EPSS 19%
Apache Kafka's code-generated protocol readers pre-size an array's collection from the element count declared by the sender, checked only against the bytes remaining in the request buffer (assuming about one byte per element). For a keyed array the reader allocates an ImplicitLinkedHashCollection of capacity (2 * count + 1), so a request of a few tens of megabytes can force a heap allocation several times larger before any element is read. The pattern affects all Kafka API types with an array field.
- CVE-2026-111073Low risk· EPSS 10%
The vulnerability affects the broker-side PushTelemetry API (KIP-714) in Apache Kafka, which validates only the compressed size of a client's metrics payload against telemetry.max.bytes before decompressing it without any limit on the decompressed output size. A small, highly compressible payload can expand to hundreds of megabytes of broker heap, causing an OutOfMemoryError.
- CVE-2026-110985Low risk· EPSS 2%
In an Apache Kafka broker using SASL/OAUTHBEARER with the default validator (DefaultJwtValidator), when sasl.oauthbearer.jwks.endpoint.url is unset and no custom verification key resolver is supplied, the validator silently falls back to ClientJwtValidator, which does not verify the JWT signature, algorithm, or issuer. The broker starts without any warning and accepts any structurally valid JWT, including "alg":"none" tokens. This is the residual of CVE-2026-33557, whose fix restored signature verification only on the path where the JWKS endpoint URL is set.
- CVE-2026-110952Low risk· EPSS 2%
An Apache Kafka broker authenticating clients via SASL/OAUTHBEARER with the bundled JWKS-backed validator always verifies the JWT signature, but does not enforce the issuer (iss) or audience (aud) claims unless sasl.oauthbearer.expected.issuer and sasl.oauthbearer.expected.audience are explicitly configured. As a result, a validly signed token from a different issuer or lacking an aud claim may be accepted. Exposure mainly affects deployments whose JWKS endpoint serves signing keys for multiple issuers or shared signing keys.
- CVE-2026-41115Medium
An improper authorization vulnerability has been identified in Apache Kafka. The CONSUMER_GROUP_DESCRIBE (69) API validates the DESCRIBE operation on the GROUP resource instead of the READ operation as documented. This can result in misconfigured ACLs and unintended security postures.
Original NVD description (English source)
Apache Kafka's code-generated readers for flexible protocol versions read a tagged-field count declared by the sender and retain a RawTaggedField object for each unknown field. That count was bounded only by the outer request frame size (socket.request.max.bytes); it was not checked against the remaining buffer bytes or capped. Because each entry can be encoded in as little as two bytes (unknown tag 0, size 0), a modest request can declare tens of millions of tagged fields and force one object allocation per declared field. Flexible request headers (header version 2, used by flexible APIs such as ApiVersions v3) are parsed this way, so the amplification applies to the request header itself. Impact: The flexible request header is parsed on the network thread before authorization, request dispatch, and request-time quota accounting. A client can send a request of a few tens of megabytes whose header declares about 20 million tagged fields, forcing roughly that many RawTaggedField allocations and exhausting broker heap or inducing process-wide GC stalls, denying broker service. Because the header is parsed for pre-authentication APIs such as ApiVersions (v3), the allocation can occur before authentication; no Kafka resource ACL or quota is required. Mitigation: Upgrade to Apache Kafka 4.2.2, 4.3.2, or 4.4.0. Until then, lowering socket.request.max.bytes and restricting network access to broker listeners reduces, but does not remove, exposure. Resolution: From Apache Kafka 4.2.2, 4.3.2, and 4.4.0: generated readers bounds-check the tagged-field count for flexible-versioned messages, rejecting a negative count, a count exceeding the remaining buffer bytes, or a count exceeding a fixed maximum (10,000), before any per-field allocation.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

