
Stay ahead in #infosec! If you're a #Linux admin or sysadmin, our newsletters are for you. Join us today! 👉 linuxsecurity.com/newsletter…

LinuxSecurity is the central voice for Linux and Open Source security news. Please visit us at linuxsecurity.com

Stay ahead in #infosec! If you're a #Linux admin or sysadmin, our newsletters are for you. Join us today! 👉 linuxsecurity.com/newsletter…

Running a daemon as a dedicated account does not automatically make command execution low impact. When an application can disclose its own configuration and credentials, manipulate its workload, and invoke OS command sinks, the service account becomes the attacker's starting security context. What happens next depends on filesystem access, capabilities, secrets, adjacent services, and local hardening. FreeSWITCH makes that boundary worth examining instead of assuming that 'non-root' means contained. **In practical terms, it is a good time to:** - inspect the FreeSWITCH systemd unit and any drop-ins with `systemctl cat freeswitch` - check the service account's UID, groups, writable directories, and access to credential-bearing configuration - review the service process and binaries with `getcap` for unexpected Linux capabilities - examine SELinux or AppArmor policy status and investigate denials that may reveal attempted boundary crossing - identify secrets readable by the FreeSWITCH account that could provide access to databases, trunks, APIs, or other hosts When you assess a service RCE, do you model the daemon's real permissions or stop at whether it runs as root? #LinuxHardening #AccessControl #LinuxSecurity #DevSecOps linuxsecurity.com/news/secur…

Browser patching is endpoint security, but it is also supply chain security. When an actively exploited browser flaw appears, downloading an installer from an arbitrary mirror to move faster creates a second problem while solving the first. Linux already gives administrators a stronger model: update through the configured, trusted repository and verify what was actually installed. **In practical terms, it is a good time to:** - verify Chrome repository definitions under `/etc/apt/sources.list.d/` or the RPM repository configuration - investigate package-signature or repository-metadata verification failures instead of bypassing them - compare installed packages with trusted repository metadata - identify locally downloaded Chrome packages that bypass normal repository management - document the approved source for emergency browser updates When patch urgency rises, does your process preserve software provenance or quietly weaken it? #SupplyChainSecurity #LinuxSecurity #OpenSource #CyberSecurity linuxsecurity.com/news/deskt…

Two clusters can both say OpenShift 5 and still have materially different Linux security baselines. An in-place upgrade from OpenShift 4.22 can remain on RHEL CoreOS 9, while a fresh OpenShift 5 installation defaults to RHEL CoreOS 10. That difference can affect kernel behavior, drivers, cryptographic defaults, compliance evidence, and hardware compatibility. Version labels at the platform layer can hide meaningful differences underneath. If inventory, vulnerability management, or compliance reporting stops at the OpenShift release, teams may compare unlike systems and assume controls are equivalent when they are not. **In practical terms, it is a good time to:** - inventory each cluster's OpenShift version and RHEL CoreOS major version together - compare node kernel versions, enabled crypto policies, and loaded modules across upgraded and newly installed clusters - validate security scanner findings against the actual node OS and vendor backport status - check compliance profiles and exceptions separately for CoreOS 9 and CoreOS 10 nodes - test required drivers and hardware integrations before treating clusters as interchangeable How many infrastructure inventories record the platform version but omit the node OS version that actually determines much of the Linux security behavior? #LinuxSecurity #OpenShift #InfrastructureSecurity #VulnerabilityManagement #Compliance linuxsecurity.com/news/cloud…

A patched kernel on disk is not the same thing as a protected kernel in memory. Kernel updates create a familiar operational gap: the package manager can report the fixed package while `uname -r` still shows the pre-update kernel because the host has not rebooted. With a mitigation-selection bug like this BHI issue, that difference is security state, not housekeeping. This is where patch compliance dashboards can overstate protection if they track installation but not activation. In practical terms, it is a good time to: - compare `uname -r` with the newest installed kernel package - identify hosts with a fixed kernel installed but an older kernel still running - verify reboot completion after kernel maintenance rather than inferring it from package success - retain running-kernel evidence with the change record When your dashboard says a kernel vulnerability is remediated, does it know what kernel is actually executing? #PatchManagement #LinuxSecurity #SecurityOperations #SysAdmin linuxsecurity.com/news/secur…

A web compromise is rarely contained by reinstalling the application that was initially exploited. If code executed through the web server account, the attacker’s practical reach depends on filesystem permissions, credentials available to PHP, writable directories, and the privileges of adjacent services. Replacing WordPress core can remove altered core files while leaving artifacts elsewhere. Recovery therefore has to follow the privilege boundary, not just the application boundary. **In practical terms, it is a good time to:** - identify directories writable by the web-server user - inspect cron jobs and systemd timers accessible to application or deployment accounts - review application configuration for database and service credentials exposed to PHP - search web-writable directories for unexpected executables or scripts - verify restored files preserve intended ownership, permissions, ACLs, and SELinux labels #LinuxSecurity #IncidentResponse #LinuxHardening #AccessControl #InfrastructureSecurity linuxsecurity.com/news/secur…

A flood of vulnerability reports can become its own availability problem for open source security. The Linux kernel already depends on maintainer review, subsystem expertise, testing, stable-tree decisions, and distribution backports. Scaling discovery without scaling trustworthy validation can shift work downstream rather than remove it. For enterprises, that means the quality of evidence attached to a finding matters almost as much as the finding count. **In practical terms, it is a good time to:** - require a minimal reproducer, affected code path, and fix reference for internally escalated kernel findings - track whether a fix is upstream, in a stable kernel, or backported by your distribution - suppress duplicate scanner findings that map to the same root cause and remediation - compare security advisories with installed package metadata before opening duplicate remediation work Better automation should reduce uncertainty for maintainers and operators, not merely increase ticket volume. #OpenSourceSecurity #Linux #VulnerabilityManagement #SecurityOperations csoonline.com/article/424526…

`ls -l` is a starting point for permission analysis, not the final answer. ACLs and security labels can change effective access without changing the familiar owner/group/other view. **In practical terms, it is a good time to:** - pair `stat` with `getfacl` on sensitive paths - use `ls -Z` where SELinux is enabled - test access as the affected user or service account Read the full analysis via the article link. #Linux #AccessControl #SecurityOperations linuxsecurity.com/howtos/lea…

A one-command environment assignment is a useful containment mechanism, but only if teams understand its scope. `NAME=value command` limits that assignment to the launched command instead of changing later commands in the parent shell. That can reduce accidental configuration persistence in administrative workflows. **In practical terms, it is a good time to:** - replace unnecessary persistent exports with command-scoped assignments - verify the target command receives only the variables it requires - inspect shell startup files for stale security-sensitive exports #Linux #SysAdmin #LinuxSecurity #DevSecOps linuxsecurity.com/howtos/lea…

If a sandbox flaw is fixed but historical file access was never logged, incident response starts with an evidence gap. You can patch the worker. You cannot retroactively prove what old jobs did without records of file, build, and storage operations. **In practical terms, it is a good time to:** - confirm file-helper and build requests are retained centrally - correlate worker activity with auditd, journald, or Kubernetes audit records - test that logs survive worker replacement and rotation #IncidentResponse #LinuxSecurity #SecurityOperations #ThreatDetection linuxsecurity.com/news/secur…

Certificate-store updates expose a dependency that is easy to miss: applications do not always trust the same certificates as the operating system. The Azure Linux release changes certificate handling tied to the eMudhra incident, but an application with its own trust store may not inherit the system-level change automatically. That turns a package update into an application-dependency question. **In practical terms, it is a good time to:** - identify applications that bundle or maintain private CA stores - compare application trust stores with the updated system trust configuration - test critical TLS connections after the trust changes are deployed - investigate certificate-chain failures before manually restoring removed trust LinuxSecurity.com examines the issue here: #LinuxSecurity #CloudSecurity #PKI #OpenSourceSecurity #InfrastructureSecurity linuxsecurity.com/news/cloud…

The hardest BusyBox problem may be finding every copy you actually operate. A patched build server does not silently repair an old container image, router firmware, rescue filesystem, or appliance image already deployed elsewhere. Those artifacts can preserve vulnerable code long after the main Linux estate looks current. This is where software inventory becomes an incident-response capability rather than a compliance exercise. If a vulnerable applet appears in an advisory, teams need to answer where that exact build exists now. **In practical terms, it is a good time to:** - search container images and extracted firmware for BusyBox binaries and symlinks - compare container image digests with approved deployment records - identify firmware and startup images produced before fixed builds entered the build pipeline - record BusyBox origin, build configuration, image digest, and deployed device population in the software inventory Can your current asset inventory trace one embedded component from build pipeline to deployed device? #SupplyChainSecurity #LinuxSecurity #DevSecOps #IncidentResponse linuxsecurity.com/news/iot-s…

A six-hour systemd timer is a refresh schedule, not proof that revocation data is current. Ubuntu 26.10 uses `upki-fetch.service` and its timer to maintain local revocation data for curl. Offline hosts, failed fetches, disabled timers, or an initial download that never completed can leave the control weaker than the configuration suggests. This is a familiar Linux security pattern: a control can be installed, enabled, and still operationally ineffective because its data pipeline failed. **In practical terms, it is a good time to:** - run `systemctl status upki-fetch.service` and inspect recent failures - run `systemctl list-timers` and confirm the upki fetch timer is scheduled and executing - inspect `journalctl -u upki-fetch.service` for failed or stale refresh attempts - verify the initial revocation-data fetch completed on newly provisioned Ubuntu 26.10 hosts - add revocation-data freshness to configuration or compliance checks where curl protects sensitive workflows If a security control depends on periodically refreshed local data, who owns detecting when that refresh quietly stops working? More detail: #Linux #LinuxSecurity #SecurityOperations #SystemAdministration #CyberResilience linuxsecurity.com/news/crypt…

Recursive cloning is convenient precisely because it hides a second trust decision. A repository can point to submodules maintained under different accounts, organizations, hosts, or security practices. Running `git clone --recurse-submodules` turns those references into immediate network retrieval without giving an administrator much time to examine what the parent project is asking for. In a mature open source security program, dependency provenance should not disappear just because Git expresses the dependency in `.gitmodules`. The commit pinned by the parent repository matters, but so do the submodule URL, ownership, access method, and whether that source is still the one your organization intended to trust. **In practical terms, it is a good time to:** - clone the parent repository without recursive submodule initialization - inspect `.gitmodules` for every declared path and URL - compare submodule hosts and owners with approved upstream sources - review the pinned submodule commits before initialization - investigate unexpected URL rewrites in system, global, and local Git configuration How many teams review submodule provenance with the same care they apply to package dependencies? More detail: #LinuxSecurity #OpenSourceSecurity #SupplyChainSecurity #Git #DevSecOps linuxsecurity.com/howtos/lea…

A green compliance dashboard can hide a vulnerable package manager if it measures activity instead of state. “Automatic updates enabled” and “APT security fix installed” are not equivalent controls. The first describes policy or automation. The second describes the actual package state on the host. That gap becomes important whenever remediation depends on a specific fixed version. This is where vulnerability management becomes an evidence problem rather than a checkbox problem. **In practical terms, it is a good time to:** - collect installed APT versions as evidence across the fleet - compare scanner findings with distribution-specific fixed versions and backport status - flag hosts whose update jobs succeeded but whose package state remains vulnerable - retain remediation evidence for systems covered by patch-management or compliance requirements Does your patch program measure successful jobs, or verified end state? #VulnerabilityManagement #Compliance #LinuxSecurity #SecurityOperations linuxsecurity.com/news/secur…

The easiest workload to recover is rarely the workload that tells you whether recovery is safe. A web server can often tolerate process recreation and dropped connections better than a database, message broker, storage service, or application maintaining significant in-memory state. That distinction matters when evaluating Kmorph’s Nginx demonstration. The useful next question is not whether another service can start. It is what happens to state that was being changed when the kernel failed. **In practical terms, it is a good time to:** - classify critical Linux workloads by statefulness and recovery semantics - test database consistency after forced host or kernel failures in a non-production environment - inspect application and storage logs for incomplete writes, replay, or recovery warnings - verify filesystem and database recovery behavior against documented RPO and RTO targets - include active I/O in resilience testing instead of testing only idle services LinuxSecurity.com examines what the current prototype has and has not demonstrated. #LinuxSecurity #SysAdmin #DataIntegrity #IncidentResponse #InfrastructureSecurity linuxsecurity.com/news/secur…