Release notes

Follow what changes in the Kabeen platform and its agents, version by version.

v3.2.1

Minor

This release fixes the silent halt in data reporting seen on some servers and removes measurement loss during brief network outages. It also brings full support for the Red Hat family: software inventory, network collection, and RPM package installation, from CentOS/RHEL 7 through recent versions.

What's new

  • The agent now reports installed-software inventory on RPM-based distributions (RHEL, Rocky Linux, AlmaLinux, CentOS): until now those servers showed no software at all.
  • Enriched network collection — DNS hostnames and UDP traffic — now works on RHEL 8 and its derivatives, including hosts still using the cgroup v1 hierarchy.

Improvements

  • The RPM installation package now installs across the whole Red Hat family, from CentOS/RHEL 7 through the most recent versions.
  • Network collection logs are easier to read: a single, explicit message when the host environment does not allow enriched collection, instead of several misleading errors.

Bug fixes

  • Fixes the issue where the agent silently stopped reporting any data after a few hours to a few days, with no visible error and while the service was still running; only a service restart brought collection back.
  • Measurements taken during a brief network outage are no longer lost: they are kept and resent once the link is back, removing the small daily gaps in metrics.

v3.2.0

Minor

This release makes process telemetry substantially lighter and improves the reliability of both network collection and the agent's secure connection. Windows Server fleets and Debian 12 / RHEL-family Linux distributions benefit directly.

What's new

  • Process monitoring now reports, for each process, a summary per 30-minute window: average and peak CPU, peak memory, and total bytes read from and written to disk.

Improvements

  • Process telemetry is around 95% lighter, reducing bandwidth usage and the agent's memory footprint on monitored servers.

Bug fixes

  • DNS hostname identification and UDP traffic counters now work on Debian 12 and RHEL-family distributions, where network collection was previously incomplete without reporting it. On the oldest Linux kernels, UDP volumes are counted without identifying the originating process.
  • On Windows servers whose root certificate store is out of date, the agent now establishes its secure connection to the Kabeen platform from the very first install — no more servers intermittently showing as offline and no more gaps in inventory. Environments relying on corporate TLS inspection continue to work exactly as before.

v3.1.0

Major

This major release introduces network activity monitoring for servers: the agent now observes and reports network flows (connections, traffic volumes, destinations, and the hostnames contacted), with reporting designed for minimal impact on servers and the network.

What's new

  • The agent now monitors server network activity: it observes and reports flows (connections, inbound and outbound traffic volumes, and the services contacted). The feature is optional and enabled in the configuration.
  • On Linux servers, the agent identifies the hostnames being contacted — not just IP addresses — and tracks both TCP and UDP traffic.
  • Installers can now be downloaded from stable “latest” links, so you can deploy the newest agent across a fleet without hard-coding a version number.

Improvements

  • Network reporting is more efficient: flows are aggregated by destination before being sent, with a window that widens over time (up to 2 hours) for long-running servers — keeping network and server overhead low while still surfacing new activity quickly after startup.
  • Network monitoring adapts to your fleet: it uses a high-performance native capture engine on recent Linux kernels and automatically falls back to a lighter method on older kernels.
  • Network capture keeps a stable, bounded memory footprint, validated by long-running endurance tests.

Bug fixes

  • On older Linux kernels where some advanced capture features cannot load, the agent now clearly reports reduced network visibility instead of appearing fully healthy — giving teams an accurate picture of what is actually being collected.

v3.0.0

Major

Version 3.0 introduces Kabeen Server Agent as the successor to Kapsul, rebuilt from the ground up to monitor Linux, macOS and Windows server fleets. This first release lays the foundation for unified, lightweight server monitoring that fits even the most demanding enterprise environments.

What's new

  • New Kabeen Server Agent that replaces Kapsul and delivers unified monitoring across Linux, macOS and Windows.
  • Automatic and transparent takeover of existing Kapsul installations: the API key and machine identity are migrated on first start, with no re-enrollment required.
  • Native support for Linux, macOS and Windows servers on both x86_64 and ARM64 architectures.
  • Installation via native .deb, .rpm, .pkg and .msi packages, with automatic registration of the systemd, launchd or Windows service.
  • Extended compatibility with enterprise Linux distributions thanks to a fully static binary that runs on older kernels and glibc, down to RHEL/CentOS 7.
  • Corporate proxy support: configuration via config.toml or through the HTTPS_PROXY, HTTP_PROXY and NO_PROXY environment variables.
  • Full hardware and system inventory: CPU, memory, disk, domain name, FQDN and operating system details.
  • Inventory of installed software on each server, including product identifier and install location.
  • Inventory of running processes and scheduled jobs.
  • Continuous collection of CPU, memory and disk metrics, with a periodic health signal confirming the agent is alive.
  • Guided MSI installer with API key entry and acceptance of the Kabeen terms of use.
  • GPG-signed Linux packages (.deb/.rpm) and Azure Trusted Signing-signed Windows installer to guarantee distribution integrity.

Improvements

  • Stronger machine identification: identity recovered from Kapsul tolerates mixed formats, preventing duplicates on the platform side.
  • Internal systemd mount points are filtered out automatically so that only the disks that matter to operators are reported.
  • Payloads sent to the backend are now compressed, reducing network usage across large fleets.
  • Clearer diagnostics on connection or proxy misconfiguration: the underlying cause is now surfaced in the logs.

Bug fixes

  • The MSI installer no longer waits on the service to reach Running state, making silent and MDM-driven deployments more reliable.