← All reports

Security News

57 items

The Hacker News

The Register

Database News

12 items

SQLServerCentral

Cloud News

14 items

Google Cloud Blog

Kubernetes Blog

Microsoft Azure Blog

Scoped Vulnerability Tracking

Database

The single most urgent item across the Database scope is CVE-2019-1068 (Microsoft SQL Server), a CISA KEV-confirmed remote code execution vulnerability with a hard remediation due date of 2026-08-29 — three days away. A KEV-confirmed entry always outranks a merely high-CVSS finding because it represents confirmed active exploitation rather than theoretical severity. This demands immediate patching or mitigation of all internet-exposed SQL Server instances ahead of the BOD 26-04 deadline. Beyond that, MongoDB's BI Connector ODBC Driver holds the highest CVSS in the dataset (9.8 CRITICAL, CVE-2026-19001), and a cluster of MariaDB Connector flaws (CVE-2026-55215, CVE-2026-55854–55860) exposes cleartext credential leakage over unencrypted transports — directly relevant to the homelab mTLS/PKI posture.

Formal Vulnerabilities

[Microsoft SQL Server] — CVE-2019-1068 (CVSS UNSCORED — CISA-confirmed active exploitation) 🔴 ACTIVELY EXPLOITED (CISA KEV)

Summary: Microsoft SQL Server contains a remote code execution vulnerability that could allow an attacker to execute code in the context of the SQL Server Database Engine service account. CISA added this to the KEV catalog on 2026-08-26 with a remediation due date of 2026-08-29. Stakeholders must apply mitigations per vendor instructions and BOD 26-04 guidance, or discontinue use of the product if mitigations are unavailable.

Affected Features / Interoperability: SQL Server Database Engine — allows code execution under the service account context. Downstream impact depends on the service account's permissions (database access, file system, network). The raw data does not specify which SQL Server versions or editions are affected.

Source:

Actions:

---

[MySQL]CVE-2026-59335CVSS 8.7 · HIGH

Summary: Improper case-sensitivity handling (CWE-178) in Cloud Foundry UAA's Identity Zone Endpoint allows a remote authenticated attacker with only `zones.write` authority to bypass the restriction that this authority does not grant access to the privileged `uaa` (system) identity zone. By referencing the zone identifier in a non-lowercase form (e.g., `UAA`), the attacker exploits a mismatch between the case-sensitive authorization layer and the case-insensitive MySQL default collation, overwriting the system zone's JWT signing key and forging admin tokens. This issue only affects UAA deployments backed by MySQL using its default collation; PostgreSQL and HSQLDB backends are not affected.

Affected Features / Interoperability: Cloud Foundry UAA identity zone authorization when backed by MySQL with default collation. Impacts JWT signing key management, zone isolation, and any Cloud Foundry deployment that trusts UAA. Replication and client drivers are not directly impacted; the compromise is at the authorization/persistence layer.

Source:

Actions:

---

[MySQL]CVE-2026-55215CVSS 7.5 · HIGH

Summary: MariaDB Connector/Node.js prior to versions 3.3.3, 3.4.6, and 3.5.3 sends credentials before completing certificate fingerprint validation when SSL is enabled without a pinned CA or server certificate. A server selecting `mysql_clear_password` as the initial authentication plugin can receive the password before the post-TLS identity check completes. An active man-in-the-middle can present a self-signed certificate, capture the database password, and authenticate directly even though the connector later rejects the server.

Affected Features / Interoperability: MariaDB Connector/Node.js TLS handshake and authentication flow. Impacts any Node.js application connecting to MariaDB/MySQL using `mysql_clear_password` auth over SSL without pinned certificates. Client drivers and connection pools that rely on this connector are affected; replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55855CVSS 6.5 · MEDIUM

Summary: MariaDB Connector/Node.js prior to versions 3.2.4, 3.3.3, 3.4.6, and 3.5.3 permits SQL injection when attacker-controlled Buffer parameters are escaped client-side under the big5, gbk, sjis, cp932, or gb18030 client character sets. The `PacketOutputStream.writeBufferEscape` method escapes bytes without the charset-aware `getMbRecognizer` logic, allowing a lead byte to consume the inserted backslash as a multibyte trail byte and leave a following quote unescaped. The default `utf8mb4` charset and the binary prepared-statement `execute` path are not affected.

Affected Features / Interoperability: MariaDB Connector/Node.js client-side escaping for multibyte character sets (big5, gbk, sjis, cp932, gb18030). Impacts applications using these charsets with Buffer parameters. Default `utf8mb4` and prepared statements are not affected. Replication and downstream integrations are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55854CVSS 5.9 · MEDIUM

Summary: MariaDB Connector/Node.js prior to versions 3.2.4, 3.3.3, 3.4.6, and 3.5.3 can disclose an account password when PAM dialog authentication is negotiated over an insecure transport. The `SendPamAuthPacketFactory` behavior for the server-side plugin dialog lacked the secure-transport gate applied to `mysql_clear_password`. With default `sslMode=DISABLE` and `restrictedAuth=null`, a hostile or on-path server can send an Authentication Switch Request for dialog over plain TCP, causing the connector to return the account password in cleartext. Properly verified TLS and a local Unix socket prevent this path; fingerprint-only server identity validation is not sufficient.

Affected Features / Interoperability: MariaDB Connector/Node.js PAM dialog authentication over unencrypted transports. Impacts connections with default SSL settings. Properly verified TLS and Unix sockets are not affected. Replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55856CVSS 5.9 · MEDIUM

Summary: MariaDB Connector/J prior to versions 2.7.14, 3.3.5, 3.4.3, and 3.5.9 can accept an untrusted self-signed certificate through the `fallbackToSystemTrustStore=true` ephemeral trust manager and record its certFingerprint for later identity binding. The initial-handshake path does not enforce the certificate fingerprint check, so `HandshakeResponse.encode()` can build and send a `mysql_clear_password` response before checking `certFingerprint`, `sslMode`, or whether the authentication plugin is MITM-resistant. An active MITM or hostile server can present a self-signed certificate, claim to be MariaDB, select `mysql_clear_password`, and receive the full database password before the connection is rejected.

Affected Features / Interoperability: MariaDB Connector/J initial TLS handshake and `mysql_clear_password` authentication. Impacts Java applications connecting with `sslMode=verify-full` or `sslMode=verify-ca` without configuring `serverSslCert` or `trustStore`. Replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55857CVSS 5.9 · MEDIUM

Summary: MariaDB Connector/J prior to versions 2.7.14, 3.3.5, 3.4.3, and 3.5.9: PAM dialog authentication can be coerced into transmitting the account password over an insecure connection. The `mysql_clear_password` plugin is gated behind a secure transport, but the sibling PAM handler `SendPamAuthPacketFactory` (named `dialog` by the server) does not declare that requirement and inherits the default `secure-required` value of false. A hostile or MITM server can send an Authentication Switch Request for `dialog` over plain TCP, causing the driver to return the user's password in cleartext when `sslMode=DISABLE` and `restrictedAuth=null` (the default configuration).

Affected Features / Interoperability: MariaDB Connector/J PAM (`dialog`) authentication over unencrypted transports. Impacts default-configuration Java applications. Properly verified TLS and local Unix sockets are not affected. Replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55858CVSS 5.9 · MEDIUM

Summary: MariaDB Connector/J prior to versions 2.7.14, 3.3.5, 3.4.3, and 3.5.9 encodes and decodes protocol text and performs client-side escaping under the assumption that the connection character set is UTF-8. A server can announce a mid-session change to `character_set_client` through OK-packet session-state tracking (including via `SET NAMES`, stored routines, triggers, server configuration, or a hostile server). If `character_set_client` changes to a non-UTF-8 value, the driver continues to read and write UTF-8 while the server interprets the same bytes under another encoding, causing silent data corruption and a charset-confusion mismatch that can defeat byte-wise quoting or escaping. The fix accepts only `utf8`, `utf8mb3`, or `utf8mb4` after initialization; any other value causes `SQLException` with SQLState `08000` and closes the connection.

Affected Features / Interoperability: MariaDB Connector/J character set handling and session-state tracking. Impacts applications where the server can change `character_set_client` mid-session. Silent data corruption and potential escaping bypass. Replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55859CVSS 5.9 · MEDIUM

Summary: MariaDB Connector/R2DBC prior to version 1.4.1 encodes and decodes all character data under the assumption that the connection character set is UTF-8. A server can announce a mid-session change to `character_set_client` through the OK-packet session-state-tracking mechanism (including via `SET NAMES` executed by a stored routine or trigger, server configuration, or a hostile/MITM server). If the new character set is not UTF-8, the driver continues to exchange UTF-8 while the server interprets the same bytes under a different encoding, causing silent data corruption and a charset-confusion mismatch that can defeat byte-wise quoting or escaping. The fix accepts only `utf8`, `utf8mb3`, or `utf8mb4` after initialization; any other value raises `R2dbcNonTransientResourceException` with SQLState `08000` and closes the connection.

Affected Features / Interoperability: MariaDB Connector/R2DBC character set handling and session-state tracking for non-blocking Java applications. Impacts reactive/database applications using R2DBC. Silent data corruption and escaping bypass. Replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

---

[MySQL]CVE-2026-55860CVSS 5.9 · MEDIUM

Summary: MariaDB Connector/R2DBC prior to version 1.4.1 does not gate clear-text password authentication plugins (`mysql_clear_password` and `dialog`/PAM) on transport encryption because the `AuthenticationPlugin` interface has no capability for a plugin to require a secure connection. A hostile or MITM MariaDB server can send an `AuthSwitchRequest` naming `mysql_clear_password` or `dialog` over a plain-TCP unencrypted connection, and `AuthenticationFlow` permits `ClearPasswordPluginFlow` or `PamPluginFlow` to return the user's password as cleartext bytes on the wire. The disclosed credentials can subsequently be used to authenticate directly to the database server.

Affected Features / Interoperability: MariaDB Connector/R2DBC clear-text authentication plugin gating. Impacts any R2DBC-based application connecting over unencrypted transports. Credential leakage enables direct database authentication. Replication and extension compatibility are not directly addressed in the raw data.

Source:

Actions:

Cloud Security

The most urgent cloud security issue is the actively exploited JFrog Artifactory path traversal vulnerability (CVE-2026-66384), which CISA has added to its Known Exploited Vulnerabilities list with a remediation deadline of 2026-09-10, allowing authenticated attackers to write outside the intended Docker cache directory and potentially compromise underlying hosts.

Formal Vulnerabilities

AWSCVE-2026-54874CVSS 7.5 · HIGH

Summary: A flaw in OpenSSL’s DTLS handling causes excessive memory buffering when a peer sends records for a future epoch during an ongoing handshake, enabling a remote attacker to trigger denial‑of‑service via memory amplification without authentication.

Affected Features / Interoperability: DTLS endpoint memory management; any service relying on OpenSSL DTLS (e.g., VPN, DTLS‑based APIs) could experience resource exhaustion leading to service unavailability.

Source:

Actions:

Google Cloud PlatformCVE-2026-19889CVSS 8.2 · HIGH

Summary: An authenticated user with Duo Agent Platform access can craft malicious model metadata in the GitLab AI Gateway to redirect model requests to an attacker‑controlled endpoint, leaking Google Vertex AI or AWS Bedrock credentials.

Affected Features / Interoperability: GitLab AI Gateway component; model request forwarding and credential storage.

Source:

Actions:

Google Cloud PlatformCVE-2026-75871CVSS 8.2 · HIGH

Summary: An authenticated user with Duo Agent Platform access can override the HTTP Host header via a crafted inline flow configuration in the GitLab AI Gateway, causing outbound model requests to be sent to an attacker‑controlled server and disclosing Google Cloud Vertex credentials and private signing keys.

Affected Features / Interoperability: GitLab AI Gateway request handling and credential management.

Source:

Actions:

KubernetesCVE-2026-54745CVSS 10.0 · CRITICAL

Summary: The Kubeflow Pipelines frontend exposes an unauthenticated SSRF via the /_proxy/ route, allowing attackers to reach internal services, extract cloud metadata credentials, and interact with Kubernetes APIs without authentication.

Affected Features / Interoperability: Kubeflow Pipelines frontend proxy middleware; internal service accessibility.

Source:

Actions:

KubernetesCVE-2026-8715CVSS 9.6 · CRITICAL

Summary: Vault Secrets Operator versions 1.3.0‑1.4.1 allow a tenant with limited Kubernetes RBAC to read arbitrary files from the operator pod’s filesystem via the AppRole authentication config and exfiltrate them, leading to privilege escalation.

Affected Features / Interoperability: Vault Secrets Operator AppRole authentication configuration; pod file system access.

Source:

Actions:

KubernetesCVE-2026-66787CVSS 8.7 · HIGH

Summary: Insufficient validation of advertised IP addresses in EndpointSlice objects within Red Hat Advanced Cluster Management’s lighthouse component lets a compromised spoke cluster inject malicious IPs, enabling man‑in‑the‑middle attacks on cross‑cluster service traffic.

Affected Features / Interoperability: Lighthouse DNS resolution of EndpointSlice objects; cross‑cluster service communication.

Source:

Actions:

KubernetesCVE-2026-55848CVSS 8.6 · HIGH

Summary: MapFish Print versions prior to 3.28.30, 3.30.32, 3.31.24, 3.33.16, and 4.0.5 fail to disable external XML entities when processing GML layer URLs, allowing unauthenticated attackers to read local files or perform SSRF via the /api/print3/print endpoint.

Affected Features / Interoperability: MapFish Print GML layer parsing; file system and outbound HTTP access.

Source:

Actions:

KubernetesCVE-2026-75889CVSS 7.7 · HIGH

Summary: Grafana Alloy’s prometheus.operator.servicemonitors component allows a user who can create or modify ServiceMonitor resources to specify an arbitrary local file via bearerTokenFile, which is then read and sent as a bearer token to an attacker‑controlled endpoint, potentially exposing the service account token.

Affected Features / Interoperability: Alloy servicemonitors component; bearer token handling and file access.

Source:

Actions:

KubernetesCVE-2026-72640CVSS 6.5 · MEDIUM

Summary: The Elastic Cloud on Kubernetes (ECK) operator accepts namespace references from secret annotations without validation, enabling a user with limited permissions to copy secrets from any namespace into a readable secret.

Affected Features / Interoperability: ECK operator secret reconciliation logic; cross‑namespace secret access.

Source:

Actions:

KubernetesCVE-2026-72648CVSS 6.5 · MEDIUM

Summary: ECK writes Elasticsearch service account tokens in cleartext into workload specifications when reconciling Fleet Server resources, allowing any principal who can read workload specs to obtain live credentials.

Affected Features / Interoperability: ECK Fleet Server reconciliation; secret handling in pod specs.

Source:

Actions:

KubernetesCVE-2026-32637UNSCORED

Summary: Velero versions before 1.18.1 allow a malicious backup tarball containing parent‑directory paths to escape the extraction directory during restore, enabling overwriting of sensitive files in the Velero pod filesystem.

Affected Features / Interoperability: Velero restore process; backup object‑storage handling and file system extraction.

Source:

Actions:

KubernetesCVE-2026-65956UNSCORED

Summary: In KubePi ≤1.6.15, SSO configuration API endpoints are exposed without admin authorization, letting low‑privileged users view or alter SSO settings, potentially leading to account takeover, and the SSO connectivity‑test can be abused as an SSRF vector.

Affected Features / Interoperability: KubePi SSO configuration API; authentication management and SSRF surface.

Source:

Actions:

Docker — CVE-2026-66384 (CVSS UNSCORED ‑‑ CISA-confirmed active exploitation) 🔴 ACTIVELY EXPLOITED (CISA KEV)

Summary: JFrog Artifactory contains a pathname validation flaw that lets an authenticated user write data outside the intended Docker cache directory under specific remote‑repository configurations, which has been observed in active exploitation. CISA lists a remediation deadline of 2026-09-10.

Affected Features / Interoperability: Artifactory Docker cache path handling; remote repository write operations.

Source:

Actions: