This commit introduces optional connection rebalancing
for TCP listeners, targeted as cases where there are a
small number of long lived connections such as service
mesh HTTP2/gRPC egress.
Part of this change involved tracking connection counts
at the per-listener level, which made it clear that we
have quite a bit of tech debt in some of our interfaces
in this area. I did various cleanups in service of this
change which leave the connection handler / accept path
in a cleaner state.
Fixes https://github.com/envoyproxy/envoy/issues/4602
Signed-off-by: Matt Klein <mklein@lyft.com>
Mirrored from https://github.com/envoyproxy/envoy @ 587e07974e6badb061ee3c9413660ab47f42750f
Description:
Risk Level: Medium (Opt-in required)
Testing: unit test, integration tested.
Docs Changes: TODO(incfly) add an architecture docs.
Release Notes: New feature, implement `Cluster.transport_socket_matches`. Envoy can be configured to use different transport socket configuration to different endpoints based on metadata match.
Fixes#8016
Signed-off-by: Jianfei Hu <jianfeih@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ cc057ce626adb2076d3dddb8dec4628c62fc5b35
This patch moves away from the paradigm of sed-style upgrading of every v2
package to v3alpha. Instead, an additional type analysis phase is
performed prior to protoxform by a protoc plugin known as the "type
whisperer".
The type whisperer produces structured type dependency information for
each .proto. The tools/type_whisperer/typedb_gen.py tool then knits
these together to provide an API-wide type dependency graph. This is
then used to determine which types need upgrading (either they have
breaking changes or transitively depend on types with such changes).
Only packages with upgraded types now undergo the v2 -> v3alpha
transition.
The API type database is checked into
source/common/config/api_type_db.pb. This may seem a strange location,
but in the future we will include the type database as a build artifact
for the Envoy binary, as it will be used by the reflection-based version
converter to find the type upgrade path for input proto.
Risk level: Low (the v3alpha protos are not used yet).
Testing: fix_format, manual inspection of diffs, bazel test //test/...,
docs build.
Part of #8082Fixes#8490
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ ad57b58cfbb256af41a467260dce2a8013b7a7fa
This allows changing the precedence of global/vhost/route level
configs when adding/removing headers.
Fixes#8450.
Signed-off-by: Raul Gutierrez Segales <rgs@pinterest.com>
Mirrored from https://github.com/envoyproxy/envoy @ 6460467c366e26d63feb1ebe1ffd9d86dcb9ceca
Adds a max request and response header count limit with default value
100. This value is configurable in the bootstrap. Runtime configurations
envoy.reloadable_features.http_connection_manager.max_request_headers_count
and
envoy.reloadable_features.http_connection_manager.max_response_headers_count
override the default limit of 100.
Both the codecs and the HCM perform this check.
Signed-off-by: Asra Ali <asraa@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 5c122a35ebd7d3f7678b0f1c9846c1e282bba079
This provides canonical BUILD formatting and puts protoxform in charge
of being able to determine import paths, without having to worry about
Bazel implications.
Part of #8082.
Risk level: Low
Testing: tools/proto_sync.py, visual inspection of diffs.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ e53f40f0e5ccc84fca5cd350416fe0f2accf8229
As part of #8082, we want to be able to (1) automatically generate BUILD
files and (2) treat packages as atomic from a "upgrade / do not upgrade"
decision perspective. This is simplified by having our BUILD targets at
package granularity, since this is what the protoxform plugin operates
on.
This PR broadens the package-level treatment that was already introduced
for Go in #8003 to Python and C++. This simplifies BUILD files
significantly and opens the way to automated generation.
There is some technical debt introduced, since all visibility controls
have been removed. This is slated for reintroduction in
https://github.com/envoyproxy/envoy/issues/8491.
As a bonus (useful for BUILD file generation), also removed the
inconsistency in BUILD package target naming for packages in envoy.api.*
and envoy.type.*. E.g. //envoy/api/v2:v2 is now //envoy/api/v2:pkg.
Risk level: Low (but this will break internal builds and require BUILD
fixups to consuming projects).
Testing: bazel test //test/... @envoy_api//...
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 4e858f17fe08224c9c089240908ccd0c518e01a7
* [#not-implemented-warn:] was barely used and its purposes are better
served by [#not-implemented-hide:].
* [#proto-status:] was there for an earlier style of versioning, where
APIs were "frozen" or "draft", etc. Now we have semantic versioning
and a regular API clock as per #6271.
Part of #8371.
Risk level: Low (docs only).
Testing: Docs rebuild.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 2c4b6f2b3d614a15f312e34f4664ebeb96d07d12
This adds a new dns_failure_refresh_rate configuration to the Cluster type to allow the configuration of the DNS refresh rate during failures (empty responses).
Risk Level: Low
Testing: Updated tests
Docs Changes: Updated docs
Release Notes: Added an entry describing the change
Fixes#7367
Signed-off-by: Venil Noronha <veniln@vmware.com>
Mirrored from https://github.com/envoyproxy/envoy @ 153bf8251be76e5eba8b04f2aeeb5d7a3a490690
Clarifies comments in discovery.proto about when response_nonce is(n't) set for (Delta)DiscoveryRequest. Also left a TODO in grpc_mux_impl_test.cc where spec-deviating behavior is expected.
Risk Level: none
Testing: comment-only change
Signed-off-by: Fred Douglas <fredlas@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ d6964e5aa985663f50759ad4cfc7f3d78d944b39
We should start tagging messages with "option deprecated = true;" if we
want them to be auto-deprecated by protoxform going forward.
Risk level: Low (v3alpha is not used yet)
Testing: bazel test @envoy_api//..., manual diff inspection.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ e3247539bcc39f3111287b479a85e51655a6f1db
This takes protoxform from being a v2 format tool to being able to
generate something similar to what ./api/migration/v3alpha.sh generates.
Risk level: Low (v3alpha not used yet)
Testing: fix_format, manual verification that the delta from
./api/migration/v3alpha.sh makes sense (for now).
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 89539d31e9fe749418d499352897c7acb474fab2
Via ./api/migration/v3alpha.sh. This picks up the changes since the last
sync, in particular the major reformat in #8309.
Risk level: Low (not used yet).
Testing: bazel build @envoy_api//...
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ c41cfbf6a33b8115a7e29a2b4a926aad4cd062be
API for #8016
Customers adopting service mesh likes mTLS ability. However, rolling it out without breaking existing traffic is hard. This is because mTLS is configured on per cluster basis. In reality, a service consists of multiple endpoints, mixed with having Envoy sidecar and without-sidecar endpoints. Client envoy can't send mTLS traffic until all server migrated to having Envoy sidecar.
This API tries to solve the issue by allowing mTLS/transport socket to configured at finer granularity, e.g. endpoint level. The endpoint has metadata label information, which will be used to decide which transport socket configuration to use from a map specified in the cluster.
So the outcome is that, xDS management server is able to configure client envoy talks to endpoints with sidecar in mTLS and plain text to endpoints without sidecar, for a single cluster.
Description:
Risk Level: N/A (API change only)
Release Notes: Cluster API change to use different transport socket based on endpoint label.
Signed-off-by: Jianfei Hu <jianfeih@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 1f7f90f7d1bdfaaab4e20198a09a52b678eab5d1
This patch introduces a new tool, protoxform, that will be the basis of
the v2 -> v3 migration tooling. It operates as a Python protoc plugin,
within the same framework as protodoc, and provides the ability to
operate on protoc AST input and generate proto output.
As a first step, the tool is applied reflexively on v2, and functions as
a formatting tool. In later patches, this will be added to
check_format/fix_format scripts and CI.
Part of #8082.
Risk level: medium (it's possible that some inadvertent wire changes
occur, if they do, this patch should be rolled back).
Testing: manual inspection of diff, bazel test //test/..., some
grep/diff scripts to ensure we haven't lost any comments.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 08b123a8321d359ea66cbbc0e2926545798dabd3
This PR adds fields to CDS that allow for configuring LRS.
Risk Level: Low
Testing: None (but if anything is needed, please let me know)
Docs Changes: Inline with API protos
Signed-off-by: Mark D. Roth <roth@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ ca3056b3aeabcfbe3750d611a8dfe44a3a7de1cf
* Move trailing detached comments that protoc ignores to some anchored place. In the near future,
this will be enforced by check_format.
* Remove comments on reserved fields; we won't be using reserved in v3 and it's a bit of a pain
preserving them.
Risk level: Low
Testing: Docs build, diff inspection, comparison against output of proto transform tool.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 22712d26c427091a47324b5f05fd0b493ee4be01
Some example regexes were rendered incorrectly on envoyproxy.io. I chose
to try to find all the example regexes and make them literal text. This
way the regexes are syntactically correct both in the raw proto and on
the doc site. Also found an H2 protocol option with a missing paragraph
due to a blank line vs. empty comment line.
Risk: low, comment-only
Testing: n/a
Docs: updated
Release Notes: n/a
Signed-off-by: Stephan Zuercher <zuercher@gmail.com>
Mirrored from https://github.com/envoyproxy/envoy @ 35b459f064ff465fc9f0f04a3f0e2adf7f85a5bc
This existed for the v1 -> v2 migration and is no longer relevant.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 358793bce7a0a3eb41f4d9961da5c61b374ceb7d
Configured via 'retriable-headers' retry policy and 'retriable_headers'
list of headers (both can be set via config or request headers) . If the
upstream response has any of the retriable headers set, retry will be
triggered.
Signed-off-by: Oleg Shaldibin <olegsh@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 81460d8482b73ba70ea15b76163a7c645e2a7b96
This change defines C++ grpc bindings for the external AuthZ interface. In the istio Security WG we're planning on using this interface to provide transparent authentication of requests. As this interface lies on the data plane we wish to implement it in C++ to maintain consistent and predictable performance
The changes allow a C++ implementation built using bazel to import Envoy as a workspace dependency and generate the C++ gRPC bindings directly without the need for complex import and generation scripts.
Signed-off-by: Nick A. Smith <nickrmc83@gmail.com>
Mirrored from https://github.com/envoyproxy/envoy @ 41932e9e6f3c932f37f77ae0a5191d65bb7ec8eb
Description: Add a new outlier detection mode which compares each host's rate of request failure to a configured fixed threshold.
Risk Level: Low
Testing: 2 new unit tests added.
Docs Changes: New mode and config options described.
Release Notes: white_check_mark
Fixes#8105
Signed-off-by: James Forcier <jforcier@grubhub.com>
Mirrored from https://github.com/envoyproxy/envoy @ 36cf26b9d6ac32f5671c24b92f9ea22ab70bc6c1
Previously, when in a panic state, requests would be routed to all
hosts. In some cases it is instead preferable to not route any requests.
Add a configuration option for zone-aware load balancers which switches
from routing to all hosts to no hosts.
Closes#7550.
Signed-off-by: James Forcier jforcier@grubhub.com
Risk Level: Low
Testing: 2 new unit tests written; manual testing
Docs Changes: Note about new configuration option added
Release Notes: added
Signed-off-by: James Forcier <jforcier@grubhub.com>
Mirrored from https://github.com/envoyproxy/envoy @ e2eb25823832d5affe602a6308c7de7ae60370bd
Remove gogoproto annotations. They can be replaced with a custom gogoproto compiler (e.g. something like https://github.com/gogo/googleapis/tree/master/protoc-gen-gogogoogleapis). I have an experimental version of it to validate that it's possible to re-apply important annotations in the compiler.
Risk Level: low
Testing: builds
Signed-off-by: Kuat Yessenov <kuat@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ e7f0b7176efdc65f96eb1697b829d1e6187f4502
This PR adds fields to CDS that allow for general-purpose LB policy configuration.
Risk Level: Low
Testing: None (but if anything is needed, please let me know)
Docs Changes: Inline with API protos
Release Notes: N/A
Signed-off-by: Mark D. Roth <roth@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ d21d92ca21cd827e8ab484244a9eefc31bfd8af4
These were missed in #8125.
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 1b3b4ae1180b67bee6395fab5c075896fb1964ec
Fixes#7982
Defines a package level proto library and its associated internal go_proto_library.
Deletes all existing api_go_proto_library, api_go_grpc_library, and go_package annotations in protos (they are not required and pollute the sources).
I deliberately avoided touching anything under udpa since it's being moved to another repository.
Risk Level: low
Testing: build completes
Signed-off-by: Kuat Yessenov <kuat@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ d504fde0ffd97017d1ddff8caa9a3b46bba9ae48
This patch establishes a v3alpha baseline API, by doing a simple copy of
v2[alpha] dirs and some sed-style heuristic fixups of BUILD dependencies
and proto package namespaces.
The objective is provide a baseline which we can compare the output from
tooling described in #8083 in later PRs, providing smaller visual diffs.
The core philosophy of the API migration is that every step will be
captured in a script (at least until the last manual steps),
api/migration/v3alpha.sh. This script will capture deterministic
migration steps, allowing v2[alpha] to continue to be updated until we
finalize v3.
There is likely to be significant changes, e.g. in addition to the work
scoped for v3, we might want to reduce the amount of API churn by
referring back to v2 protos where it makes sense. This will be done via
tooling in later PRs.
Part of #8083.
Risk level: Low
Testing: build @envoy_api//...
Signed-off-by: Harvey Tuch <htuch@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 085d72b490c124a02849812798f5513a8df9ae72
* api:Add a flag to disable overprovisioning in ClusterLoadAssignment
Signed-off-by: Jie Chen <jiechen@google.com>
* api:Add [#next-major-version and [#not-implemented-hide to the comment
for field of disable_overprovisioning in ClusterLoadAssignment
Signed-off-by: Jie Chen <jiechen@google.com>
* api:Refine comments for the new added bool flag as suggested.
Signed-off-by: Jie Chen <jiechen@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ b28edcae3da369ed9969796f89118c1f898a27d9
Description: add Scoped RDS routing logic into HCM. Changes include:
* in ActiveStream constructor latch a ScopedConfig impl to the activeStream if SRDS is enabled
* in the beginning of ActiveStream::decodeHeaders(headers, end_stream), get routeConfig from latched ScopedConfig impl.
This PR is the 3rd in the srds impl PR chain: [#7704, #7451, this].
Risk Level: Medium
Testing: unit test and integration tests.
Release Notes: Add scoped RDS routing support into HCM.
Signed-off-by: Xin Zhuang <stevenzzz@google.com>
Mirrored from https://github.com/envoyproxy/envoy @ 7960564746700312b4fec21711c4387794f5ba06
The java_outer_classname is unintentionally duplicated in the new
udp_listener_config and regex proto files. This changes them to unique
names that match the predominant naming scheme.
Signed-off-by: Bryce Anderson <banderson@twitter.com>
Mirrored from https://github.com/envoyproxy/envoy @ f80188ebc4b592754449c740f831123b4de41ab3
The libstdc++ std::regex implementation is not safe in all cases
for user provided input. This change deprecates the used of std::regex
in all user facing paths and introduces a new safe regex matcher with
an explicitly configurable engine, right now limited to Google's re2
regex engine. This is not a drop in replacement for std::regex as all
language features are not supported. As such we will go through a
deprecation period for the old regex engine.
Fixes https://github.com/envoyproxy/envoy/issues/7728
Signed-off-by: Matt Klein <mklein@lyft.com>
Mirrored from https://github.com/envoyproxy/envoy @ eff020170c6267e6c8dc235473f7fc85c5b1e07d
Depending on the execution environment in which envoy is being run, it
is possible that some of the assumption on the clock are maybe not
holding as previously commented. With some sandboxing technologies the
clock does not reference the machine boot time but the sandbox boot
time. This invalidates the assumtpion that the first update in the
cluster_manager will most likely fall out of the windows and ends up
showing a non intuitive behavior difficult to catch.
This PR simply adds a comment that will allow the reader to consider
this option while reading to the code.
Signed-off-by: Flavio Crisciani <f.crisciani@gmail.com>
Mirrored from https://github.com/envoyproxy/envoy @ 640b5a436d2ce8e637d28225d5b4f0aae307dede
This PR adds BoringSSL private key API abstraction, as discussed in #6248. All comments and discussion is welcomed to get the API sufficient for most private key API tasks.
The PR contains the proposed API and the way how it can be used from ssl_socket.h. Also there is some code showing how the PrivateKeyMethodProvider is coming from TLS certificate config. Two example private key method providers are included in the tests.
Description: tls: support BoringSSL private key async functionality
Risk Level: medium
Testing: two basic private key provider implementation
Docs Changes: TLS arch doc, cert.proto doc
Signed-off-by: Ismo Puustinen <ismo.puustinen@intel.com>
Mirrored from https://github.com/envoyproxy/envoy @ 9a3a234c3cd1f7b6707746d84eb74414b1990c73
data-plane-api(CircleCI)