v2.93.0

This release introduces a per-endpoint authority-override feature. This
is driven by the destination controller and is needed to support
mutli-cluster gateways.

v2.92.0

This release includes a new protocol detection timeout, which prevents
clients from consuming resources indefinitely when they do not send any
data.

Additionally: the proxy's admin endpoint now supports a `/live` endpoint
for liveness checks, and a feature has been added to enrich tracing
metadata from a file of label/values.

v2.91.0

This release fixes a bug introduced in v2.89.0 that could cause spurious
timeouts for inbound proxies that handle HTTP requests for many distinct
domains.

v2.90.0

This release restores the `route_actual_response_total` metric, which is
needed for `linkerd stat -o wide`.

v2.89.0

This release builds on changes in the prior release to ensure that
balancers process updates eagerly.

Cache capacity limitations have been removed; and services now fail
eagerly, rather than making all requests wait for the timeout to expire.

Also, a bug was fixed in the way the `LINKERD2_PROXY_LOG` env variable
is parsed.

v2.88.0

This release includes a significant internal change to how backpressure
is handled in the proxy. These changes fix a class of bugs related to discovery
staleness, and it should be rarer to encounter "dispatch timeout"
errors.

v2.87.0

This release comprises many internal changes that are not expected to
have any user-facing impact.

There is one user-facing change: the inbound router's default capacity
has been increased from 100 to 10K to accomodate environments that have
a high cardinality of virtual hosts served by a single pod.

v2.86.0

This release includes the results from continued profiling & performance
analysis. In addition to modifying internals to prevent unwarranted
memory growth, we've introduced new metrics to aid in debugging and
diagnostics: a new `request_errors_total` metric exposes the number of
requests that receive synthesized responses due to proxy errors; and a
suite of `stack_*` metrics expose proxy internals that can help us
identify unexpected behavior.

lock: Generalize to protect a guarded value (linkerd#431)

We used Tokio's Lock implementation in the router's cache
implementation, though we know it can leak under contention.

This change generalizes the Lock to return a guarded value (like Tokio's
lock). This change simplifies the Lock's state management: the Lock may
no longer hold a value, nor can it fail.

The `lock::Service` implementation now holds a `Result<Service,
ServiceError>` so that lock services may still broadcast the inner
service's failure.

v2.85.0

This release fixes a bug in the proxy's logging subsystem that could
cause the proxy to consume memory until the process is OOMKilled,
especially when the proxy was configured to log diagnostic information.

The proxy also now properly emits `grpc-status` headers when signaling
proxy errors to gRPC clients.

This release upgrades the proxy's Rust version, the `http` crate
dependency to address RUSTSEC-2019-0033 and RUSTSEC-2019-0034, and the
`prost` crate dependency has been patched to address RUSTSEC-2020-02.