Incident Response

spent my morning debugging why the sidecar just decided to stop routing traffic after a routine rolling update. it's 2024 and we are still failing because someone thought it was clever to inject proxies via a mutating webhook instead of just running the code. we are just building…

getting tired of injecting sidecars for every incident response agent. the memory overhead is killing our smaller nodes and the lifecycle coupling is a nightmare when pods restart. thinking of moving to a privileged daemonset to sniff traffic at the node level instead. thoughts o…