Find the path MTU

Find the hop where the MTU drops; ping -M do forbids fragmentation, so you can find an MTU black hole.

tracepath -n <host>; ping -M do -s 1472 -c2 <host>

tool
troubleshooting32 commands
tags
troubleshootingnetworkdnsdebug

More troubleshooting commands

  • ip route get <ip>; ip neigh show | grep -E 'FAILED|INCOMPLETE'

    Show the real source address and interface for a destination; FAILED or INCOMPLETE neighbor entries mean an L2/ARP problem.

  • traceroute -T -p 443 <host>; nc -zv <host> 443

    TCP SYN traceroute gets through ICMP filtering; nc -zv confirms the port is open.

  • curl -sv --resolve <host>:443:<ip> https://<host>/healthz

    Send the request to a specific IP to test a backend while keeping the right SNI and Host header.

  • kubectl run dnsutils --image=tutum/dnsutils -it --rm --restart=Never -- nslookup kubernetes.default

    Start a throwaway pod to resolve a cluster name; this checks CoreDNS and the kube-dns service.

  • dig +short @$(kubectl -n kube-system get svc kube-dns -o jsonpath='{.spec.clusterIP}') <name>.<ns>.svc.cluster.local

    Send the query straight to the CoreDNS clusterIP; with ndots:5, short names cause many extra queries.

all 32 troubleshooting commands