TLDR Tech / DevOps
Auto-diagnosing Kubernetes alerts with HolmesGPT and CNCF tools
A two-person SRE team at STCLab cut alert investigation time from 15-20 minutes to under 2 minutes by deploying HolmesGPT with custom runbooks that reduced wasted tool calls from 16 to 2 per investigation. The team found that markdown runbooks specifying which tools to skip per namespace mattered more than model selection, with the same model scoring 4.6 out of 5 with runbooks versus 3.6 without. It now handles about 12 unique daily investigations at roughly $12 per month.
Read the full story