Oliver Gould

> * When a pod gets SIGTERM, it stops accepting new connections but allows in-flight connections to progress > * Clients may not yet know the pod is terminating and...

To put it another way: we want to allow for some grace period between a kubelet announcing its intention to kill a container and our decision to stop allowing connections....

It's probably helpful to be clear about some assumptions. I assume: * The vast majority of traffic is short-lived messages (e.g., HTTP req/rsp completing in

@UsingCoding Thanks for pointing this out. The proxy already supports `LINKERD2_PROXY_INBOUND_ACCEPT_KEEPALIVE`, `LINKERD2_PROXY_INBOUND_CONNECT_KEEPALIVE`, `LINKERD2_PROXY_OUTBOUND_ACCEPT_KEEPALIVE, LINKERD2_PROXY_OUTBOUND_CONNECT_KEEPALIVE`. That is, we support configuring independent keeplalives for the inbound server, inbound client, outbound server, and...

It sounds like this is more generally asking for gRPC-specific metrics. As far as I understand, it wouldn't be possible for Linkerd to measure this sort of thing arbitrarily at...

> Is the nit about the partitioning of the route types into discrete submodules? Or does the nit just mean that this the preferred style / convention for linkerd crates?...

I'm +0 on renaming the existing tests/mod.rs (sgtm, tioli).

The audit CI failure is expected, but it looks like this is still depending on a version of the gateway bindings with `timeout` -- I think a `cargo update -p...

Hi @GrigoriyMikhalkin. Thanks for taking the time to submit this and sorry for the delay and getting you a review. I'm curious about a slightly modified approach that might lessen...

This was fixed in https://github.com/linkerd/linkerd2/releases/tag/edge-24.3.4. Please let us know if issues persist.