Quentin Machu

Hi @jicki, Yeah, it's just a timing issue. In those new logs with TTL=1min, the unhealthy member gets removed at 3:40:15, and new rejoins at 03:40:25 - which is fine....

Thanks for the detailed suggestion. NLBs are good, better than ELBs for the use-case obviously for some of the reasons you stated above, but also a few others regarding flow...

Apps that forcefully log to disk are a major pain when using containers.

I would also love to have this as it'd make the completion actually be able to use fzf's full potential. It is actually discussed a bit in https://github.com/Aloxaf/fzf-tab/issues/65 where @romkatv...

Not having individual PV per pod makes the whole operator worthless.

Also because you are relying on creating individual pods (with no anti-affinity even), rather than a statefulset - you can't actually even create PVCs/PVs that would carefully match the scheduling...

This is borderline ridiculous.

Same issue, it's pretty bizarre. 

Experiencing the same issue here. 5s delays on every, single, DNS lookup, 100% of the time. Similarly, `insert_failed` does increase for each DNS query. The AAAA query, that happens a...

@dcowden Kubernetes 1.10.1, Container Linux 1688.5.3-1758.0.0, AWS VPCs, Weave 2.3.0, kube-proxy IPVS. My guess is that it depends how fast/stable your network is?