linux

  • An incident with no engineer

    An outage at 04:00 in August, resolved before anybody woke up. A worker crash loop that would have been a two-hour outage in 2023.

  • A second host, and the first thing that was not replicated

    Everything in the provisioning script arrived on the new host, and three things that were not in it did not. A provisioning script describes configuration and a host…

  • The second host, and the failover we finally built

    Seven years on one host, and a business requirement that finally arrived. A recovery time that assumes the host comes back.

  • Debian 13, and a provisioning script that needed nothing

    A major distribution upgrade, by rebuild, with no changes to the provisioning script at all. A provisioning script that is exercised monthly is a script that works when…

  • The shell script that is still a shell script

    A deploy script rewritten twice — into a compiled language and back — and stable for eighteen months at a hundred and eighty lines. Shell is the right…

  • A systemd unit with a restart limit that finally fired

    A burst limit configured in 2023 as a precaution, which took three years to be needed. Without the burst limit the worker would have restarted every five seconds…

  • A restart limit that fired for the first time

    A systemd burst limit configured in 2023 as a precaution, which took three years to be needed. Without the limit the worker would have restarted every five seconds…

  • The service running since 2020 that nothing depended on

    A systemd unit enabled and running for five years, consuming a gigabyte, serving a feature that was removed in 2021. Nothing alerts on a service that is working…

  • An ssh certificate authority, one year on

    A year of eight-hour certificates rather than an authorized_keys file, and the operational reality of it. The mid-incident expiry is the failure mode that theory does not predict,…

  • An internal certificate from a private authority, expiring

    A service-to-service certificate issued by an internal authority in 2022, with a three-year validity nobody had recorded. A certificate with a three-year validity is a reminder set three…