2.2.0-Preview.1
Pre-release
Pre-release
MihaZupan
released this
17 May 17:50
·
59 commits
to main
since this release
This release supports .NET 6.0 and .NET 8.0. See Getting Started.
The Yarp.ReverseProxy packages are available on NuGet.org.
Breaking changes
HeaderMatchMode.NotExists
now also matches headers with an empty value (#2412 by @qianyuansun).- This makes
NotExists
exactly the opposite ofExists
. Previously, empty values would not match either mode.
- This makes
- Errors triggered by the client disconnecting are now logged under a
Debug
severity (#2503 by @MihaZupan).ForwarderError.RequestCanceled
,RequestBodyCanceled
, andUpgradeRequestCanceled
are now logged under a newILogger
event with aDebug
severity. Other errors continue to be logged under the existing event with aWarning
severity.- How errors are mapped to different
ForwarderError
s remains the same.
Other changes
- Added support for Kestrel Start/Stop events on
IKestrelTelemetryConsumer
(#2360 by @ajay-sainy). - Added support for controlling
OutputCachePolicy
in the route configuration (#2328 by @witskeeper).- See the Output Caching documentation for more details.
X-Forwarded-For
now prefers the IPv4 form for dual-mode sockets (#2400 by @hacst).- E.g.
127.0.0.1
is used instead of::ffff:127.0.0.1
.
- E.g.
- Fixed a bug where active health checks would consider new destinations as
Healthy
instead ofUnknown
(#2415 by @robbieknuth).- Destinations now remain as
Unknown
until enough health probes have been sent.
- Destinations now remain as
- Added a new
Query
property to active health check configuration (#2421 by @robbieknuth).- Previously, only
Path
could have been specified, and?
would be escaped to%3F
.
- Previously, only
- Added support to
HttpSysDelegator
for detaching from and re-initializing queues (#2426 by @NGloreous). - Fixed a race condition reliability bug in
HttpSysDelegator
(#2487 by @NGloreous). OperationCanceledException
s thrown by response transforms as part of error handling are now ignored (#2452 by @MihaZupan).- Fixed a bug where YARP validation of
UseRequestTimeouts
was too strict (#2501 by @MihaZupan). - Reduced the performance overhead of updating the activity timeout timer (#2475 by @MihaZupan).
- Many documentation improvements.
New Contributors
- @larsbj1988 made their first contribution in #2293
- @Bambi-pa-hal-is made their first contribution in #2350
- @ajay-sainy made their first contribution in #2360
- @joegoldman2 made their first contribution in #2381
- @hacst made their first contribution in #2400
- @allantargino made their first contribution in #2411
- @robbieknuth made their first contribution in #2415
- @qianyuansun made their first contribution in #2412
- @jsedlak-microsoft made their first contribution in #2436
- @jamesmontemagno made their first contribution in #2433
- @eltociear made their first contribution in #2456
For a full list of changes see here.