TOPICS

linux

systems, desktop and server administration

note The journal eats gigabytes before anyone notices The default journal cap is a tenth of the partition. On a small root that means gigabytes of logs nobody wants. systemd · journal · disk 09/2026 project Infrastructure stabilization Stabilizing an organization’s technical backbone: administration rules, documentation, a platform for testing and hosting, and a link to the infrastructure one level up. documentation · system-inventory · security · volunteers 07/2026 active project mojeWebFoundry A framework for building several separate static sites on SvelteKit. It came out of moving away from Jekyll and doubles as a way to find out what current tooling can do. sveltekit · static-site · ci · ai · monorepo 08/2026 active link How Complex Systems Fail The classic text by R. Cook – why failure never has a single cause. Required reading for anyone who runs systems. essay · systems 12/2025 note Ubuntu 25.04 (Plucky Puffin): TPM disk encryption, an ARM64 desktop and APT 3.0 A non-LTS release from April 2025: TPM-backed disk encryption in the installer, an official ARM64 desktop, kernel 6.14 with sched_ext, APT 3.0 and the end of apt-key. ubuntu · release · tpm · apt 04/2025 explainer SELinux from the kernel to the policy: labels, modules and finding bugs How SELinux hangs off the LSM hooks, what the context on every file means, how a module is built from .te and .if, and how a policy bug is actually found. selinux · lsm · policy 11/2024 explainer DISA STIG: what it is and how severity is read What DISA STIGs are, how the CAT I/II/III severity categories are read, and why rules are not applied blindly across the board. stig · compliance · security · hardening 02/2023 note OpenSSH 9.0: scp over SFTP and hybrid key exchange An April 2022 release in which scp stopped using the legacy protocol and the default key exchange gained a post-quantum component. openssh · scp · sftp 04/2022 note Podman: containers without a daemon and without root A container tool with no permanently running daemon that can run as an ordinary user and generate systemd units. podman · containers · systemd 02/2022 project SELinux development in an enterprise distribution Working on SELinux in an enterprise Linux distribution: policies, their bugs and passing the knowledge on to colleagues. selinux · policies · lsm · training 01/2022 archived link openSUSE MicroOS: a system that does not update while running An immutable system with transactional updates on btrfs snapshots, where applications live in containers rather than in the root filesystem. opensuse · microos · btrfs 11/2021 note The end of CentOS 8 and the distributions that replaced it After the December 2020 announcement that CentOS 8 would end, two independent RHEL rebuilds appeared during 2021 – AlmaLinux and Rocky Linux. centos · rhel · almalinux · rocky-linux 06/2021 project Security certification against DISA STIG Preparing a minimal enterprise distribution for DISA STIG certification: architecture, test adaptation, mitigations and judging which rules apply to the system at all. stig · compliance · hardening · sle-micro 05/2021 archived note WireGuard in the mainline Linux kernel The WireGuard VPN protocol landed directly in the mainline Linux kernel in March 2020, without needing a separately maintained module. wireguard · vpn · kernel 03/2020 project Infrastructure as Code with Ansible Idempotent infrastructure deployment with Ansible – system and applications, roles driven by host groups, and contributions to an OpenStack deployment. ansible · iac · openstack · deployment · idempotence · hardening 01/2020 archived explainer AppArmor: an introduction to MAC for Linux applications What AppArmor is, how mandatory access control through profiles works, and how profiles are created and controlled. apparmor · security · linux-kernel · profiles 06/2019 project Cloud platform test automation A framework for automated testing of OpenStack cloud platform updates, storage integration and CI/CD – including a period leading the test team. openstack · test-automation · ci-cd · storage · qa 06/2019 archived project A model railway controlled from Kubernetes A model railway fully controlled and monitored from a small Intel NUC computer via Arduino, with the applications running in a Kubernetes cluster. kubernetes · intel-nuc · arduino · monitoring · talk 01/2019 archived explainer AIX for a Linux admin: where Linux habits stop working The differences a Linux administrator meets when moving to AIX on IBM Power – tooling, LVM, LPARs, enterprise storage and backups, compared with SLES and RHEL. aix · ibm-power · lvm · lpar · sles · rhel 10/2018 explainer Why and how enterprise distribution maintenance packages get tested Why a fix package for an enterprise Linux distribution goes through testing before release – regression testing, multiple architectures and virtualization. qa · regression-testing · packages · virtualization 01/2018 explainer Kernel isolation: PV, HVM and OS-level virtualization in Linux The difference between paravirtualization, full hardware virtualization and OS-level virtualization (containers) in Linux. virtualization · containers · xen · kvm · namespaces 10/2017 note Debian 9 (Stretch): MariaDB instead of MySQL, X without root A June 2017 release in which MariaDB became the default database and the X server stopped requiring root privileges. debian · mariadb · xorg 06/2017 guide Cacti for a small virtualized infrastructure How to deploy Cacti on a dedicated host and use SNMP to watch load on both the hosts and the virtual machines – paravirtualized and fully virtualized alike. cacti · snmp · monitoring · xen 04/2017 guide ProxyJump: through a bastion without holding tunnels in your head Getting scp, rsync and git through a jump host by writing it into ssh_config once. ssh · bastion 03/2017 note Removing a package while breaking dependencies Force-removing a single package on a Debian-based system without letting apt pull the rest of a dependency chain with it. dpkg · apt · debian 04/2015 guide A read-only root on Raspbian that survives a cold power-off How to turn Raspbian on an SD card into a read-only root so the device survives being unplugged, and how to turn it into an image for mass deployment. raspberry-pi · raspbian · read-only · tmpfs 04/2015 explainer High availability and shared storage for a small XEN cluster How high availability and shared storage work for a small cluster of virtual machines on XEN. xen · high-availability · nas · san · virtualization 03/2015 project Read-only Raspberry Pi for daily menu displays Daily menu displays in canteens across the country: a Raspberry Pi that survives being unplugged and deploys without anyone touching it on site. raspberry-pi · raspbian · read-only · digital-signage · deployment 03/2015 archived guide Moving off a probably compromised server onto clean infrastructure How the move of an organization’s services off a probably compromised server onto new infrastructure actually went: the old stack, the split into virtual machines, the upgrade trouble and the hardening. security · kvm · migration · hardening · php · containers 01/2015 project Update testing across architectures Verifying fixes of an enterprise Linux distribution before release – on x86, IBM Z and PowerPC, with a focus on virtualization. qa · regression-testing · virtualization · kvm · ibm-z 05/2014 archived project Infrastructure migration Moving an organization’s services off a probably compromised server onto a newly designed infrastructure, running it long term and leading a technical team of volunteers. migration · virtualization · security · e-mail · jekyll · volunteers 04/2014 archived project Packaging for RPM and Debian and the Open Build Service Building packages for both RPM- and Debian-based distributions – spec, dsc and control files, builds locally and in the Open Build Service. rpm · deb · open-build-service · packages 01/2014 active note Maildir: login mail notification and reading with mailx Getting new-mail notifications on login (including SSH) and reading local mail with mailx when the delivery agent uses Maildir. maildir · mailx · pam · debian 12/2013 project XEN virtualization infrastructure Consolidating several servers onto the XEN XCP virtualization platform with high availability, shared storage and backups – including migration of running services. xen · virtualization · high-availability · nas · backup · migration 11/2013 archived note XEN / XCP: a handful of useful notes Scattered fixes from running XEN Cloud Platform: VNC keymap, LVM storage type, the HVC console and CPU hotplug in a guest. xen · xcp · virtualization 05/2013 guide Examining a RAID member from a disk image with dd_rescue and mdadm Imaging a failing drive with dd_rescue and inspecting its mdadm RAID member over a loop device – without touching the original disk. dd-rescue · mdadm · loop-device · raid 03/2013 guide Restoring deleted files with extundelete Recovering deleted files from an ext3/ext4 filesystem with extundelete – mount read-only first and work by inode. extundelete · ext4 · recovery 12/2012 project AIX and Linux administration Installation, upgrades, configuration and support of AIX and Linux systems on IBM Power – logical partitions, SAN, backups and high availability. aix · powerha · lpar · san · ksh 01/2012 archived note Debian 6.0 (Squeeze): dependency-based boot ordering A February 2011 release that switched the default boot script ordering to dependency-based sequencing and offered a FreeBSD kernel port as a technology preview. debian · sysvinit · insserv 02/2011 troubleshooting Portmapper hangs at boot or shutdown The portmapper service stalls for a minute during boot or halt – how the issue was tracked down and resolved. portmap · init · rpc 06/2010 project Network administration for companies and the public sector End-to-end administration of networks and servers for private companies and the public sector – from network devices to mail, VPN and storage. network-administration · openvpn · iscsi · firewall · active-directory 04/2010 archived project Infrastructure monitoring Watching load and availability of servers and networks – from SNMP and an information system for monitoring and helpdesk to Cacti on top of a virtualization infrastructure. monitoring · cacti · snmp · availability · helpdesk 01/2010 archived