Denial of service CVEs.
Published CVEs classified as denial of service, ranked by exploit odds.
-
CVE-2026-118899
When using the listen directive with the ssl parameter in the http block with TLSv1.3 enabled, an attacker can use a specially crafted TLS connection to cause an infinite loop in the NGINX worker process while reading the HTTP request body. As a result, the worker process stops accepting new connections and processing existing connections. Note: Software versions which have reached End of Technical Support (EoTS) are not evaluated. Impact: This vulnerability allows a remote, unauthenticated attacker to cause a denial-of-service (DoS) of NGINX. There is no control plane exposure; this is a data plane issue only. Note: Software versions which have reached End of Technical Support (EoTS) are not evaluated.
-
CVE-2026-68539
fulgur converts untrusted HTML/CSS into PDF, commonly on a server that processes input supplied by many tenants. From version 0.4.4 to before version 0.29.0, the CSS string-set named-string values resolved from a document were accumulated per element with no memory budget, letting a small crafted document amplify into large retained memory (and the CPU to copy it) — results in a denial of service. This issue has been patched in version 0.29.0.
-
CVE-2026-86696
Allocation of Resources Without Limits or Throttling vulnerability in esaml in dropbox esaml allows an unauthenticated attacker to exhaust the Erlang VM atom table and crash the node via unsigned LogoutResponse messages. esaml:status_code_map/1 in src/esaml.erl maps any StatusCode value it does not recognize to an atom with list_to_atom/1 on the last colon-separated segment of the URN. esaml_sp:validate_logout_response/2 verifies a signature only when the message carries one, so an unsigned LogoutResponse posted to the single-logout endpoint reaches esaml:decode_logout_response/1 and mints one permanent atom per distinct status value. Atoms are never reclaimed and the VM caps the atom table at about one million entries, so an attacker who posts that many distinct values aborts the node. This issue affects esaml: from 0.1.0 onward.
-
CVE-2026-93471
Improper Handling of Highly Compressed Data (Data Amplification) in esaml_binding in dropbox esaml allows an unauthenticated attacker to exhaust server memory and crash the Erlang VM via a crafted compressed SAMLResponse. esaml_binding:decode_response/2 base64-decodes the attacker-supplied SAMLResponse, inflates it with zlib:unzip/1 with no bound on the output, then converts the result to a charlist with binary_to_list/1 before parsing. This runs on the SAML consume (ACS) and single-logout endpoints, which accept unauthenticated requests by design, and it runs before any signature check. A small deflate stream reaches a compression ratio near 1000 to 1, so a request body of tens of kilobytes inflates to hundreds of megabytes and then to several gigabytes of list heap, so a few requests exhaust memory and take the node down. This issue affects esaml: from 0.1.0 onward.
-
CVE-2026-115825
HTTP::Server::PSGI versions before 1.0056 for Perl allow a single request to crash the server by declaring a large Content-Length. The request body is read with sysread() calls sized by the remaining Content-Length. Perl grows the read buffer to the requested length before any data arrives, so the server tries to allocate as much memory as the client declares. A remote attacker can send one request with a large Content-Length and a few body bytes, and perl exits with an out-of-memory error before the application is called. HTTP::Server::PSGI runs as a single process, so the server stays down until restarted.
-
CVE-2026-126659
BlueprintUE Self-Hosted Edition 1.0.0 before 4.3.5 contains a resource exhaustion vulnerability that allows unauthenticated attackers to fill server storage by repeatedly creating anonymous blueprints. Attackers can send large requests to HomeController::doProcessCreateBlueprint() with a 'never' expiration, bypassing purges to fill the disk and deny service.
-
CVE-2026-128129
AIOHTTP is an asynchronous HTTP client/server framework for asyncio and Python. Prior to 3.14.4, aiohttp buffers all additional connection data received while a request marked for upgrade, such as a WebSocket handshake, is being handled, without a size limit or flow control. A remote attacker can stream data behind an upgrade request to a handler that does not immediately accept or reject the upgrade and cause unbounded memory consumption. This issue is fixed in version 3.14.4.
-
CVE-2026-122096
Uncontrolled Resource Consumption vulnerability in the sketch deserialization of Apache DataSketches Java. Several deserializers trusted sizes and counts declared in the serialized input. A small crafted input could make an application allocate an excessive amount of memory, leading to an OutOfMemoryError. A crafted string passed to FrequentLongsSketch.getInstance(String) could also make deserialization, or a later update, loop indefinitely. Only applications that deserialize sketches from untrusted sources are affected. This issue affects Apache DataSketches Java: through 9.0.0. Users are recommended to upgrade to version 9.1.0, which fixes the issue.
-
CVE-2026-120367
Allocation of Resources Without Limits or Throttling vulnerability in Apache Axis2/Java. The JSON message builders, including the Gson, Moshi and HTTP/2 variants, buffered the entire request body with no size limit, so an unauthenticated caller could exhaust server memory with a large JSON request. Version 2.0.2 bounds JSON request bodies with the configurable jsonMaxRequestSize parameter. This issue affects Apache Axis2/Java: through 2.0.1. Users are recommended to upgrade to version 2.0.2, which fixes the issue.
-
CVE-2026-119052
llama.cpp b2877 before b11935 contains an uncaught exception vulnerability in the ggml-rpc server that allows unauthenticated remote attackers to crash the rpc-server by sending oversized message lengths. After the HELLO handshake, attackers can send a command such as SET_TENSOR with a length above std::vector::max_size(), triggering an uncaught std::length_error that aborts the process and denies service to all clients.
-
CVE-2026-123865
An unauthenticated HTTP/2 client can permanently leak Node.js server memory and file descriptors. If a peer triggers a protocol error, for example an oversized frame header, while the session is already shutting down, nghttp2 tears the session down internally but Node.js never notices, leaving a dangling session: the socket stays in CLOSE_WAIT and all of Node.js's HTTP/2 state for that connection is held in memory indefinitely, with no further cost to the attacker. The attacker can start the shutdown remotely by sending GOAWAY(NO_ERROR), so it is reliable against default HTTP/2 servers. It is an incomplete-fix case of CVE-2026-48937: that fix detected nghttp2 internal terminations by watching for GOAWAY sent while not shutting down, but missed terminations that arrive while the session is already shutting down. This vulnerability affects Node.js 26.x, 24.x, and 22.x.
-
CVE-2026-68540
fulgur converts untrusted HTML/CSS into PDF, commonly on a server that processes input supplied by many tenants. From version 0.6.0 to before version 0.33.0, quadratic style-list reprocessing during containing-block resolution for absolutely-positioned pseudo-elements cause denial of service. This issue has been patched in version 0.33.0.
-
CVE-2026-115448
Uncontrolled recursion in Apache ZooKeeper. The server serializes its data tree to a snapshot by recursing one stack frame per znode level, with no depth bound. A client permitted to create znodes (the default root ACL is open) can build a chain of nested znodes deep enough that taking a snapshot exhausts the JVM stack and kills the snapshot thread; sizes stay within jute.maxbuffer, so existing limits do not prevent it. This is a denial of service that becomes persistent: on restart the server rebuilds the deep tree from its transaction log and fails again while taking the startup snapshot, so it does not return to service, and recovery can entail data loss. The same log affects a leader and its followers, so it can spread across an ensemble. This issue affects Apache ZooKeeper 3.8.0 through 3.8.7 and 3.9.0 through 3.9.6. Users are recommended to upgrade to 3.8.8 or 3.9.7, which fix the issue. As a workaround, use ACLs and authentication to restrict znode creation rather than relying on the default open root ACL.
-
CVE-2026-108529
Inefficient Algorithmic Complexity vulnerability in the HTTP/2 implementation of mtrudel bandit allows an unauthenticated remote attacker to exhaust server CPU via small SETTINGS or connection-level WINDOW_UPDATE frames. The attacker opens thousands of streams whose responses block on the connection send window, then sends cheap frames that each make the server walk the whole queue of blocked sends. In the reporter's run, with 3,000 blocked streams a single 9-byte SETTINGS frame took 133 ms to process. do_pending_sends/2 in lib/bandit/http2/connection.ex retries every queued send and, before each retry, removes that entry from the queue with List.delete/2, so one frame costs O(n²) list operations for n blocked streams. max_concurrent_streams defaults to :infinity, so the attacker chooses n. This issue affects bandit: from 0.3.4 before 1.14.0.
-
CVE-2026-110499
Allocation of Resources Without Limits or Throttling vulnerability in the HTTP/2 implementation of mtrudel bandit allows an unauthenticated remote attacker to grow per-connection memory and CPU cost without bound via a large SETTINGS_HEADER_TABLE_SIZE. Bandit resizes its HPACK encoder table to whatever size the client advertises, up to 2^32-1 bytes, and indexes response headers into it. Because responses carry distinct values such as date or content-length, nothing is evicted and the table grows with every response for the life of the connection, while each lookup scans it linearly. The SETTINGS handling in Bandit.HTTP2.Connection (lib/bandit/http2/connection.ex) passes remote_settings.header_table_size to HPAX.resize/2 with no server-side cap, and encode_headers encodes every response header as :store. Serving N responses on one connection therefore costs O(N²) CPU in the connection process. This issue affects bandit: from 0.3.6 before 1.14.0.
-
CVE-2026-109584
Improper Handling of Highly Compressed Data (Data Amplification) vulnerability in the HTTP/2 implementation of mtrudel bandit allows an unauthenticated remote attacker to exhaust server memory via a small HPACK-compressed HEADERS block. The block stores one large cookie value in the HPACK dynamic table and then references it tens of thousands of times with 1-byte indexed fields. In the reporter's run, a 49 KB block made a server with default settings allocate a 183 MB cookie binary before any application code ran, and every concurrent stream can repeat the cost. Bandit.HTTP2.Connection caps only the compressed block (max_header_block_size), and its check of the decoded list against max_header_list_size never fires because that setting defaults to :infinity. combine_cookie_crumbs/1 in lib/bandit/http2/stream.ex then joins every crumb into one binary. This issue affects bandit: from 0.3.3 before 1.14.0.
-
CVE-2026-111101
Allocation of Resources Without Limits or Throttling vulnerability in the HTTP/1.1 chunked body reader of mtrudel bandit allows an unauthenticated remote attacker to exhaust server memory and CPU via a request body made of many 1-byte chunks. The chunked reader in lib/bandit/http1/socket.ex keeps every chunk as its own list cell and binary term until IO.iodata_to_binary/1 joins them, and nothing limits the number of chunks. The :length option of read_body/2 counts only data bytes, so tiny chunks cost far more memory than their payload: 48 MB of 1-byte chunks grew one handler process to about 500 MB and took about 6 s of CPU. The route does not need to read the body, because on keep-alive connections ensure_completed/1 drains unread bodies through the same reader. This issue affects bandit: from 0.1.0 before 1.14.0.
-
CVE-2026-114213
Allocation of Resources Without Limits or Throttling vulnerability in the WebSocket fragment reassembly of mtrudel bandit allows an unauthenticated remote attacker to exhaust server memory via a fragmented message made of many 1-byte continuation frames. Bandit.WebSocket.Connection checks max_fragmented_message_size against the payload bytes received, but keeps every non-final continuation frame as a new list cell around its own binary, and nothing limits the number of fragments. A 1-byte fragment costs 7 bytes on the wire but about 150 to 170 bytes of server memory, so one connection can hold about 150 times the configured limit, roughly 1.2 to 1.4 GB at the default 8 MB. Ping frames between fragments keep the connection alive past the idle timeout. This issue affects bandit: from 1.11.0 before 1.14.0.
-
CVE-2026-108525
Allocation of Resources Without Limits or Throttling vulnerability in the HTTP/2 implementation of mtrudel bandit allows an unauthenticated remote attacker to exhaust server memory via DATA frames sent past the advertised receive window. The attacker requests a response larger than the 65,535-byte initial window and withholds WINDOW_UPDATE, so the stream process blocks while sending and stops reading, then floods small DATA frames on that stream. Every frame is queued in the stream's mailbox until the BEAM node runs out of memory. Bandit.HTTP2.Connection.handle_frame/3 in lib/bandit/http2/connection.ex forwards each DATA frame to the stream process without consulting the receive window, and Bandit.HTTP2.FlowControl.compute_recv_window/2 replenishes the window as data arrives rather than as it is consumed, and never answers a violation with FLOW_CONTROL_ERROR. This issue affects bandit: from 0.3.1 before 1.14.0.
-
CVE-2026-68538
fulgur converts untrusted HTML/CSS into PDF, commonly on a server that processes input supplied by many tenants. From version 0.15.0 to before version 0.27.0, resolving the CSS counters(name, sep) function in generated content was bounded neither in the size of a single expansion nor in the aggregate work across an element tree. An attacker able to submit HTML/CSS to a fulgur-based conversion service can trigger CPU and memory exhaustion with a trivially small payload, denying service to the host and any co-tenants. This issue has been patched in version 0.27.0.
-
CVE-2026-111108
Apache Kafka's code-generated protocol readers pre-size an array's collection from the element count declared by the sender. That count is checked only against the bytes remaining in the request buffer, which assumes about one byte per element; because most elements need several bytes, a request can declare far more elements than the buffer holds and still pass. For a keyed array the reader then 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 is emitted into every generated reader, so it affects all Kafka API types with an array field. Impact: Request parsing happens before any authorization check. On a listener that does not require authentication (for example PLAINTEXT), an unauthenticated attacker can send a few concurrent requests that each declare a very large= array length and allocate many times their wire size in broker heap, crashing the broker with an OutOfMemoryError. The reporter demonstrated three concurrent ~90 MB ProduceRequest (v3) connections exhausting a 1 GB heap (~8x amplification). On listeners that require authentication, the attacker needs valid credentials but no specific ACL, since parsing precedes authorization.. 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 cap the initial array pre-allocation at MessageUtil.MAX_PREALLOCATED_ARRAY_CAPACITY (1000) and grow it on demand, and reject a declared array length above MessageUtil.MAX_ARRAY_LENGTH (1,000,000).
-
CVE-2026-126623
Inefficient algorithmic complexity in the JSON reader of Progressive Robot hMailServer 6.3.6 and 6.3.7 allows a remote attacker to hold the server's mail delivery threads or its REST API's worker threads, making mail delivery, the REST API and the webmail unavailable. 6.3.6 fixed CVE-2026-107574, in which reading a JSON object searched the member names already read for each new name, by keeping the names in a hash set; the set uses the standard library's hash, which takes no secret, so member names chosen to fall in one bucket of it still cost time growing with the square of their number. Against the released reader, one such object of 16 MB took about 5 seconds to read in the Linux build, and one of under 6 MB about 23 seconds in the Windows build, where objects of the same sizes with other names took under a tenth of a second. A remote unauthenticated attacker reaches the reader by mailing such an object as a TLS-RPT report, up to 16 MB once decompressed, to a hosted domain's report mailbox, where an administrator has set one; it is read on a mail delivery thread (ten by default), and those threads deliver all mail, local and outbound, so a stream of such reports stops all delivery for as long as it continues. It is also reached without an account through the reply to a password-protected message, with the message's link alone, where the REST API is on (it is off by default) and a public base URL for such links is set; and by any account through the webmail's routes granted 16 MB, and by an application's API key through the send API. Each such request holds one of the REST API's four worker threads, in the process that also serves SMTP, IMAP and POP3.
-
CVE-2026-97200
Allocation of Resources Without Limits or Throttling vulnerability in hackney_conn in benoitc hackney allows a malicious HTTP server to exhaust client memory via an oversized or highly compressed response body. hackney:request/5 and hackney:body/1 return the whole response body, and nothing bounds how much the connection process buffers. Any server the client requests can trigger it. hackney_conn:read_full_body/2 and the HTTP/2 and HTTP/3 accumulators append every chunk until the body ends. With {auto_decompress, true} the buffered body is then inflated in one step with zlib:gunzip/1 or zlib:inflate/2, which return the whole output at once. This issue affects hackney: from 2.0.0-beta.1 before 4.9.0.
-
CVE-2026-111115
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.
-
CVE-2026-123770
Allocation of Resources Without Limits or Throttling vulnerability in Apache Axis2/Java. The multipart/form-data and application/x-www-form-urlencoded message builders read the request body directly from the transport stream with no size limit, so the servlet container's post-size limit did not apply, and multipart temporary files were not deleted. An unauthenticated caller could exhaust memory or disk. Version 2.0.2 bounds these bodies with configurable limits and deletes temporary files. This issue affects Apache Axis2/Java: through 2.0.1. Users are recommended to upgrade to version 2.0.2, which fixes the issue.
-
CVE-2026-120135
Uncontrolled Recursion vulnerability in Apache Thrift Java bindings. This issue affects Apache Thrift: before 0.24.0. Users are recommended to upgrade to version 0.24.0, which fixes the issue.
-
CVE-2026-114349
When a broker decodes a compressed record batch, the per-record decoder reads a declared body size from the decompressed stream and allocates a buffer of that size before validating it against the bytes available or any configured limit. The size is attacker-controlled, and the batch CRC does not constrain it, so a small (around 90-byte), CRC-valid compressed produce request can declare a body size of up to about 2 GiB and force the broker to attempt that allocation, raising OutOfMemoryError. The record size limit (max.message.bytes), the request size limit, and client/broker quotas bound only the compressed wire size, not the declared decompressed size. Impact: An authenticated principal with Write on a topic can crash broker request-handler threads (reached when a batch's codec differs from the topic's compression.type and records are re-encoded) and any consumer that fetches the record, denying service. On brokers started with -XX:+ExitOnOutOfMemoryError, a common hardening, a single crafted request terminates the broker JVM. More seriously, a crafted record is accepted at ingest and written to disk; when the log-cleaner later reads it on a compacted topic, the OutOfMemoryError (an Error, not an Exception) terminates the cleaner thread, which is not restarted. With the default of a single cleaner thread, log compaction then stops for the entire broker, including the internal __consumer_offsets and __transaction_state topics, and the condition persists across restarts until the offending record is removed. The impact is availability only; no data is disclosed or altered. Mitigation: Upgrade to Apache Kafka 4.2.2, 4.3.2, or 4.4.0, and set max.decompressed.message.bytes (at the broker and/or topic level) to a bounded value appropriate for your workload. Upgrading alone is not sufficient: for backward compatibility this configuration defaults to an effectively unlimited value (about 2 GiB), so a broker remains exposed until the limit is set. Choose a value at or slightly above the largest legitimate decompressed record your applications produce, and comfortably below the broker heap divided by the number of request-handler and cleaner threads. Until you can upgrade and configure the limit, restrict Write access to trusted producers; quotas and max.message.bytes do not mitigate this issue. Resolution: From Apache Kafka 4.2.2, 4.3.2, and 4.4.0: the broker validates a record's declared decompressed body size against max.decompressed.message.bytes before allocating, on the produce-validation, log-compaction, and timestamp-lookup paths, and rejects oversized or negative sizes with InvalidRecordException. Operators must set this configuration to a bounded value to be protected, as its default is effectively unlimited.
-
CVE-2026-77622
Sliver is a command and control framework. Prior to 1.7.6, an HTTP(S) listener using the PNG encoder accepts attacker-controlled PNG dimensions without an effective decoded-size or pixel limit in util/encoders/images.go, while the listener nonce space implemented in server/encoders/encoders.go is small enough to search remotely. An unauthenticated attacker who identifies a valid nonce can submit a highly compressed image to PNGEncoder.Decode, where it expands during png.Decode, exhausting server memory and causing a complete loss of availability. This vulnerability is fixed in 1.7.6.
-
CVE-2026-17279
IBM API Connect 10.0.8.0 through 10.0.8.12 and 12.1.0.0 through 12.1.1.2 could allow a remote authenticated attacker to cause a denial of service due to uncontrolled resource consumption. Certain requests can cause internal processing to stall, which could make the service unavailable.
-
CVE-2026-105270
Allocation of Resources Without Limits or Throttling vulnerability in esaml_binding in dropbox esaml allows an unauthenticated attacker to exhaust the Erlang VM atom table and crash the node via a crafted SAMLResponse. esaml_binding:decode_response/2 parses the decoded SAMLResponse with xmerl_scan:string/2, which interns every element name, attribute name, and namespace prefix it reads as a permanent atom. The SAML consume (ACS) and single-logout endpoints call it on unauthenticated requests before any signature check, so one well-formed document with thousands of distinct names mints thousands of atoms while staying inside the 128 KB request body limit. Atoms are never reclaimed and the VM caps the atom table at about one million entries, so a few dozen requests abort the node. This issue affects esaml: from 0.1.0 onward.
-
CVE-2026-100302
Allocation of Resources Without Limits or Throttling in the multiRead request handler of Apache ZooKeeper allows a remote attacker to crash the server (denial of service) with a single small request. The multiRead handler places no limit on the number of bundled read operations or on the cumulative size of the response it builds in memory. Because each operation's result is fully materialized and retained before serialization, a request that is small on the wire — within the jute.maxbuffer limit — but that references a large znode many times can force the server to allocate far more heap than the request size implies, exhausting the JVM heap and terminating the server with an OutOfMemoryError. Neither jute.maxbuffer nor largeRequestMaxBytes prevents this, as both bound only the inbound request, not the response. This issue affects Apache ZooKeeper from 3.7.0 through 3.9.6. Users are recommended to upgrade to version 3.9.7, which enforces configurable limits on the multiRead operation count and cumulative response size.
-
CVE-2026-113873
SmileCDR HAPI FHIR does not properly handle highly compressed data, allowing an unauthenticated attacker to send compressed payloads that exhaust server heap or crash certain services when decompressed.
-
CVE-2026-117677
When parsing an uploaded Helm chart, Gitea decompressed the archive and read every tar entry while looking for `Chart.yaml` without limiting the decompressed size, and did so before the package size quota was checked; the quota only considers the compressed upload size. A user with package write access to an owner, for example their own account on an instance with open registration, could repeatedly upload small, highly compressed archives that make the server spend CPU time proportional to the decompressed size, degrading the availability of the Gitea instance. The issue does not allow code execution or data access.
-
CVE-2026-123863
A reachable assertion in Node.js HTTP/2 session teardown lets any unauthenticated client abort the server, over both TLS and cleartext. A client can arrange for an Http2Stream that still holds nghttp2-allocated header buffers to outlive its session: the stream is removed from the session's stream set but never detached, so when ~Http2Session() runs the memory accounting is nonzero and the CHECK_EQ(current_nghttp2_memory_, 0) assertion fails, terminating the process. The trigger needs only a default http2 server whose handler responds before the request body has ended, which is typical. It is a variant of CVE-2024-27983 that the earlier fix did not cover, because that fix only detached streams still tracked in the session. This vulnerability affects Node.js 26.x (since 26.9.0) and 24.x (since 24.22.0).
-
CVE-2026-115817
Starman versions before 0.4019 for Perl allow a stalled request body to hold a worker indefinitely, as read_timeout covers only the headers. Content-Length and chunked request bodies are read with blocking reads and no deadline. A client that sends complete headers and then stalls or trickles the body keeps the worker waiting until it disconnects. A remote attacker can take every worker with a few such connections, five with the default worker count, so the server stops answering requests. Version 0.4019 applies read_timeout to each wait for body data and adds a body_timeout limit, 300 seconds by default, on the total time to read a body.
-
CVE-2026-127218
Axigen Mail Server 10.3.3.67 through 10.3.3.x, 10.4.42 through 10.4.x, 10.5.29 through 10.5.x, 10.6.x before 10.6.41, and 10.7.x before 10.7.3 contains an unauthenticated denial of service vulnerability caused by uncontrolled recursion when validating query parameter and cookie values with a recursive regular expression engine before authentication in the WebAdmin and WebMail services. Attackers can send a single HTTP request with a query parameter or cookie value of a few thousand characters to exhaust the worker thread stack and crash the server process, taking down SMTP, IMAP, POP3, and web services repeatedly.
-
CVE-2026-119762
Axigen Mail Server before 10.5.69, 10.6.x before 10.6.40, and 10.7.x before 10.7.2 contains an unauthenticated denial of service vulnerability caused by unbounded nesting depth when parsing SOAP/XML and JSON request bodies before authentication. Attackers can send a single deeply nested request to WebMail, WebAdmin, or WebMail Proxy listeners to exhaust the stack and crash SMTP, IMAP, POP3, and web services repeatedly.
-
CVE-2026-115776
Twisted is an event-based framework for internet applications, supporting Python 3.6+. Prior to 26.10.0rc1, twisted.words.protocols.irc.IRC.dataReceived retains all bytes received without a line feed in its per-connection buffer and applies no maximum line length. The twisted.words.service.IRCUser server subclass inherits this behavior, and buffering occurs before PASS, NICK, or USER authentication commands. An unauthenticated connection can continuously grow the buffer until the shared server process runs out of memory and drops all users. This issue is fixed in version 26.10.0rc1.
-
CVE-2026-118761
Kubeflow Pipelines enables users to build and deploy portable, scalable machine learning workflows. Prior to 2.18.0, the API server decompressed uploaded pipeline archives without adequately bounding expanded data while locating or reading the pipeline file. A small ZIP or gzip-compressed tar archive can consume disproportionately large memory and exhaust API-server resources. Multipart pipeline and pipeline-version upload processing performs decompression before namespace authorization, so a reachable caller does not need a successful pipeline-create authorization decision. URL-import handlers authorize before fetching and decompressing a remote archive. The amount of expansion required to interrupt service depends on deployment resource limits and concurrent activity, and no specific archive is guaranteed to crash every deployment. This issue is fixed in version 2.18.0.
-
CVE-2026-116078
Allocation of Resources Without Limits or Throttling vulnerability in the HTTP/2 request body reader of mtrudel bandit allows an unauthenticated remote attacker to exhaust server memory via a request body sent as many 1-byte DATA frames. do_read_data/5 in lib/bandit/http2/stream.ex keeps every DATA frame payload as its own list element until the read returns. The :length option of read_body/2 counts body bytes, not frames, so a body of 1-byte frames costs a list cell and a binary term per byte. Interleaving the frames across up to 500 concurrent streams per connection keeps each stream consuming its frames as they arrive; in the reporter's test, 100 streams held 25 to 40 MiB of stream process memory for 5 MB sent. Only routes whose Plug reads the request body are affected, which includes any route behind Plug.Parsers. This issue affects bandit: from 0.3.1 before 1.14.1.
-
CVE-2026-124279
Allocation of Resources Without Limits or Throttling vulnerability in elixir-grpc grpc_server allows unauthenticated attackers to exhaust server memory via client-streaming or bidirectional-streaming requests. The streaming read clause of GRPC.Server.Adapters.Cowboy.Handler.info/3 (grpc_server/lib/grpc/server/adapters/cowboy/handler.ex) asks Cowboy for up to 8,000,000 bytes per read and compares the running total with max_body_size only after the read returns. The gRPC length prefix of a message is not checked before the message is buffered. Each stream can therefore buffer up to one full read beyond the configured limit before it is refused, and many concurrent streams exhaust node memory. Releases before 1.0.6 apply no limit to streaming requests at all. This issue affects grpc_server: from 1.0.0 before 1.1.0.
-
CVE-2026-100300
Improper handling of exceptional conditions in the quota subsystem of Apache ZooKeeper allows an attacker able to write under the /zookeeper/quota subtree — unauthenticated on default deployments, where that subtree is world-writable — to crash the server's critical request-processing thread and render the on-disk data directory permanently unbootable. This issue affects Apache ZooKeeper up to and including 3.9.6. Users are recommended to upgrade to version 3.8.8 or 3.9.7, which fixes the issue.
-
CVE-2026-119723
When parsing request variables such as query-string and form parameters, HHVM did not check that appending a nested array succeeded before descending into it. A remote, unauthenticated attacker could send a GET or POST request which would crash the HHVM server process before any application code runs, leading to denial of service. This issue affected HHVM prior to 23c061f4e804a51badd5f85227e65185651253b5.
-
CVE-2026-120828
A replica-set database server node does not limit or reclaim internal wait records created while handling a client-supplied read option that names a point in time the node will never reach. A party with network access and no credentials can cause these records to accumulate without bound and persist after the requests end, which may make the server run out of memory and become unavailable.
-
CVE-2026-123868
An unauthenticated HTTP/2 client can exhaust Node.js server memory because inbound DATA queued on an undrained stream is charged to no session budget. Each inbound DATA frame creates one JavaScript Buffer view in the receiving stream's readable queue, but nothing charges that queue to current_session_memory_: the socket read buffer is accounted only during parsing and released immediately, so whatever stays queued is accounted to nothing. A peer that fills a stream's receive window with one-byte DATA frames pays about 10 wire bytes per frame while the server retains about 114 bytes per frame, so a single stream on one connection can retain hundreds of megabytes from tens of megabytes sent. maxSessionMemory neither bounds the growth nor refuses new streams, because Http2Session::CanAddStream gates admission on the same counter that never sees the queue. This vulnerability affects Node.js 26.x and 24.x.
-
CVE-2026-130091
A caller with write permission in a namespace can terminate temporal-proxy through memory exhaustion by storing highly compressed payloads in Workflow history, and the stored history causes the same failure on every later read. temporal-proxy 1.2.0 decompresses every payload marked as compressed in the responses it forwards, whether or not compression is configured. Each payload is limited to 4 MiB after decompression, but the total for a response is not, and the decoded payloads are held in memory together. The caller can store many small, highly compressed payloads through the proxy or directly at the upstream Temporal service. When that history is read through the proxy, for example by a Worker processing a task, the proxy can run out of memory and be terminated. The codec server's /decode endpoint is affected in the same way when it is enabled. Only temporal-proxy 1.2.0 is affected; earlier versions do not decompress payloads. The issue is fixed in 1.2.1. Upgrade to 1.2.1 or later.
-
CVE-2026-126613
Allocation of resources without limits in the JSON reader of Progressive Robot hMailServer 6.3.2 through 6.3.7 allows a remote attacker to exhaust the server's memory and hold its threads, making mail delivery, the REST API and the webmail unavailable. The reader built a tree of values far larger than the text it read, copying each value into every level above it, and set no bound on the number of values, so a single JSON document of many small values, within the 16 MB the server reads, raised the service's memory by about 2 GB and held one thread for seconds to minutes. From 6.3.4 a remote unauthenticated attacker reaches the reader by mailing such a document as a TLS-RPT report, up to 16 MB once decompressed, to a hosted domain's report mailbox, where an administrator has set one; it is read on a mail delivery thread (ten by default), and those threads deliver all mail, local and outbound. From 6.3.4 it is also reached without an account through the reply to a password-protected message, with the message's link alone, where the REST API is on (it is off by default) and a public base URL for such links is set; and by any account, from 6.3.2 through the webmail's send and draft routes and from 6.3.4 through its other routes granted 16 MB and the send API; the webmail's send, draft and settings routes parsed the body again for each member read. The REST API serves every request on four worker threads, in the process that also serves SMTP, IMAP and POP3, so a few such documents at once make it unavailable and can exhaust the machine's memory and end the whole service. This is a separate defect from CVE-2026-107574, the reader's handling of duplicate member names, fixed in 6.3.6.
-
CVE-2026-125321
pgEdge Postgres MCP Server enables SQL queries against PostgreSQL databases through Model Context Protocol compatible clients. Prior to 1.1.0, deployments exposing the nondefault HTTP transport with authentication enabled used the leftmost caller-supplied X-Forwarded-For value as the client address for login rate limiting. ExtractIPAddress in internal/auth/middleware.go supplied that value to IsAllowed and RecordFailedAttempt in internal/auth/rate_limiter.go, allowing an unauthenticated caller to rotate addresses to evade throttling or assert another user's address to consume that user's limiter bucket. The published nginx configuration appended rather than replaced X-Forwarded-For, making the issue reachable without an operator misconfiguration. Username-keyed account lockout did not use the client address, but shipped defaults disabled that separate control through MaxFailedAttemptsBeforeLockout: 0. This issue is fixed in version 1.1.0.
-
CVE-2026-112255
Initialization of a resource with an insecure default vulnerability in OpenText Advanced Authentication Server allows Excessive Allocation. This issue affects Advanced Authentication Server: from 6.4.0.0 through 6.4.3.x, from 6.5.0.0 through 6.5.0.x, and from 6.6.0.0 through 6.6.1.0. The fix for this is issue is to create a secure configuration. The 6.6.2.0 release has a best practice entry if the configuration is unsecure. A secure configuration is to NOT allow creation/recreation of Unauthenticated Endpoints. If creation/recreation of Unauthenticated Endpoints is required, you should only allow it from trusted networks. To remedy this issue, change your configuration to NOT allow creation/recreation of Unauthenticated Endpoints. If you require Unauthenticated Endpoints, you MUST change your configuration to use a non-empty whitelist to specify the trusted networks. An empty whitelist will allow all networks to create/recreate endpoints.
-
CVE-2026-108949
Denial of Service via OOM vulnerability in Apache ActiveMQ, ActiveMQ Broker. An authenticated OpenWire client that sends a ConsumerInfo for a temporary queue/topic owned by another connection is rejected with InvalidDestinationException, but the rejected Subscription is left behind in the region's subscriptions map. Repeating the request leaks broker heap until OOM because the broker does not clean up the rejected subscriptions in the map. This issue affects Apache ActiveMQ: from 6.0.0 before 6.3.3, before 5.19.12; ActiveMQ Broker: from 6.0.0 before 6.3.3, before 5.19.12. Users are recommended to upgrade to version 6.4.0, 6.3.3, or 5.19.12, which fixes the issue.
A list of CVEs whose flaw type is denial of service, sorted by how likely they are to be exploited (EPSS). Classification uses the CWE weakness id where present, and a small local model otherwise.