code
code
> This makes sense but can we just do the implementation in this PR? > > > > /wait yeah， I can do it. But I may need to support...
> I didn't realize that the cookie hash policy required modifying response headers. It's the `Generated` mode of cookie hash policy. If the client didn't provide the specificied cookie, the...
> Another possible approach here would be that instead of configuring the hash policy as part of the LB policy config, we could just add a hash policy field to...
I initially want to use the `envoy.type.v3.HashPolicy` in the LB. But it's too simple and cannot meet our requirement: **to replace and finally deprecate the route level hash policy.** Then...
hi, matt, now, everything is here, ready for a review :) cc @mattklein123
> I don't think we want the router filter to have this built-in. I think we would want some new upstream filter to do this. If we build this into...
> Sorry @wbpcode @markdroth but I don't have the bandwidth to due justice to understanding the disagreements on this one. I would find another person who has the time to...
> LGTM from me as well. However I would like to understand the need to go from chrono::seconds to uint64_t. This looks like a regression in readability without any gain....