resolvedFriday deploy went to production
Root cause: confidence. Fix: rolled back, apologised to the pager, wrote a rule. Status: rule ignored the following Friday.
Friends call me SejiL. I keep computers running so other people can break them in exciting new ways. A résumé would list achievements. This lists incidents.
Root cause: confidence. Fix: rolled back, apologised to the pager, wrote a rule. Status: rule ignored the following Friday.
Brain entered read-only mode. Emails from this window should be treated as drafts written by a friendly stranger. Failover to tea succeeded at 60% capacity.
Spent four hours ruling out everything except DNS. It was DNS. It is always DNS. A haiku has been added to the runbook.
Window regularly overrun by "one more kubectl get pods". Duration under review. Review keeps getting postponed.
Found 14 idle things nobody remembered creating, named test-final-v2-DONOTDELETE. Deleted them. Nobody noticed. That was the point.
$ kubectl describe human sejil Name: sajjad (alias: SejiL) Namespace: earth Labels: role=devops habitat=terminal humor=dry Image: coffee:latest # never pin, live dangerously Status: Running Restarts: 1 (every morning) Limits: patience: high, unless you say "works on my machine" meetings: 2/day, then OOMKilled Liveness: http-get /snacks every 90m Tolerations: yaml-indentation:NoSchedule legacy-bash:PreferNot Events: Warning BackOff asked "quick question?" at 17:59 Normal Pulled fresh idea from the shower Normal Scaled side projects 3 → 7
Escalation policy is real. Responsiveness is not guaranteed.
Resilience is only proven by breaking things. Go on. It self-heals. Probably.