- Ya
- 17 Mei 2026, 17:05 UTC
- Mwandishi
- Kamo
- Ahadi ya
- bb9a0fc
Default RKE2 max-pods=110 was being hit, blocking kamolos rolling-update deploys (maxSurge:1,maxUnavailable:0 needs slot to create new pod before killing old; deadlocked at the cap). k1m1 has 48 cores + 125 GiB RAM and a /24 pod CIDR (254 IPs/node), so 200 fits cleanly: ~100 GiB projected for 200 avg-sized pods stays under the memory ceiling and leaves ~54 IPs of CIDR headroom. Applied directly to /etc/rancher/rke2/config.yaml on k1m1 and rke2-server restarted; this YAML is the desired-state record. Not applied via kubectl because RKE2 reads from disk on the node — see the comments in the file for the manual re-apply procedure.