Re:Service provider story about tracking down TCP RSTs
by spydum on Saturday September 01, 2018 @07:19PM
(#17894425)
i tend to agree: anycasting stateful services is always a gamble.. however, i do think including ttl in the ecmp hashing alg is prone to failure, and is probably a poor default.
[ Parent
| on HN ]
Re:Service provider story about tracking down TCP RSTs
by alecbenzer on Sunday September 02, 2018 @12:39PM
(#17897550)
You mean 'stateful' at the IP layer, right? So anything using TCP counts as "stateful"?
Like, an application which implemented "stateless" HTTP endpoints would still run into this issue?