fix sometimes returns response body to HEAD requests #3985

zyfjeff · 2018-07-29T01:57:25Z

Signed-off-by: tianqian.zyf tianqian.zyf@alibaba-inc.com

Description: fix sometimes returns response body to HEAD requests
Risk Level: low
Testing: unittests and integration test
Docs Changes:
Release Notes:
[Optional Fixes #Issue] #3932
[Optional Deprecated:]

mattklein123

Thanks for working on this. High level comment to get started.

mattklein123 · 2018-07-30T04:24:55Z

include/envoy/http/filter.h

@@ -203,7 +203,8 @@ class StreamDecoderFilterCallbacks : public virtual StreamFilterCallbacks {
   *                       response headers.
   */
  virtual void sendLocalReply(Code response_code, const std::string& body_text,
-                              std::function<void(HeaderMap& headers)> modify_headers) PURE;
+                              std::function<void(HeaderMap& headers)> modify_headers,
+                              bool is_head_request) PURE;


High level comment: Does this need to be part of the API? Would it be better if the connection manager kept track of this and did the right thing?

You are right, I will refactor

mattklein123

Thanks for working on this. In general the change looks correct, but it needs tests for all the different permutations you did this for:

AsyncClient
conn_manager_impl (early response and filter response)

mattklein123 · 2018-07-30T17:02:40Z

source/common/http/async_client_impl.cc

@@ -112,6 +112,10 @@ void AsyncStreamImpl::encodeTrailers(HeaderMapPtr&& trailers) {
 }

 void AsyncStreamImpl::sendHeaders(HeaderMap& headers, bool end_stream) {
+  if (headers.Method() &&


Method should always be set so I don't think we need to check

mattklein123 · 2018-07-30T17:28:24Z

source/common/http/conn_manager_impl.cc

@@ -481,6 +481,11 @@ const Network::Connection* ConnectionManagerImpl::ActiveStream::connection() {

 void ConnectionManagerImpl::ActiveStream::decodeHeaders(HeaderMapPtr&& headers, bool end_stream) {
  request_headers_ = std::move(headers);
+  if (request_headers_->Method() &&


Method can't be null

mattklein123 · 2018-07-30T17:28:44Z

source/common/http/conn_manager_impl.h

@@ -188,6 +188,10 @@ class ConnectionManagerImpl : Logger::Loggable<Logger::Id::http>,
    // called here may change the content type, so we must check it before the call.
    FilterHeadersStatus decodeHeaders(HeaderMap& headers, bool end_stream) {
      is_grpc_request_ = Grpc::Common::hasGrpcContentType(headers);
+      if (headers.Method() &&


Method can't be null

HttpConnectionManagerImplTest.*have many test, there is no set method header, I'm going to fix it

mattklein123 · 2018-07-30T17:29:04Z

source/common/http/conn_manager_impl.h

@@ -396,6 +402,7 @@ class ConnectionManagerImpl : Logger::Loggable<Logger::Id::http>,
    // By default, we will assume there are no 100-Continue headers. If encode100ContinueHeaders
    // is ever called, this is set to true so commonContinue resumes processing the 100-Continue.
    bool has_continue_headers_{};
+    bool is_head_request_{false};


do we need this at both levels for early response before we have filters?

mattklein123 · 2018-07-30T17:29:20Z

source/extensions/filters/http/fault/fault_filter.cc

@@ -75,6 +75,7 @@ Http::FilterHeadersStatus FaultFilter::decodeHeaders(Http::HeaderMap& headers, b
  // faults. In other words, runtime is supported only when faults are
  // configured at the filter level.
  fault_settings_ = config_->settings();
+


revert unrelated change

mattklein123 · 2018-07-30T17:29:24Z

source/extensions/filters/http/grpc_web/grpc_web_filter.cc

@@ -45,7 +45,6 @@ Http::FilterHeadersStatus GrpcWebFilter::decodeHeaders(Http::HeaderMap& headers,
    return Http::FilterHeadersStatus::Continue;
  }
  is_grpc_web_request_ = true;
-


revert unrelated change

Sorry, forgot to check before submission

I think you have a few whitespace removals you can revert

mattklein123 · 2018-07-30T17:29:32Z

source/extensions/filters/http/jwt_authn/filter.cc

@@ -1,5 +1,6 @@
 #include "extensions/filters/http/jwt_authn/filter.h"

+#include "common/http/headers.h"


revert all changes in file

mattklein123 · 2018-07-30T17:29:38Z

source/extensions/filters/http/jwt_authn/filter.h

@@ -43,6 +43,7 @@ class Filter : public Http::StreamDecoderFilter,
  State state_ = Init;
  // Mark if request has been stopped.
  bool stopped_ = false;
+  bool is_head_request_{false};


mattklein123 · 2018-07-30T17:29:44Z

source/extensions/filters/http/rbac/rbac_filter.cc

@@ -1,5 +1,6 @@
 #include "extensions/filters/http/rbac/rbac_filter.h"

+#include "common/http/headers.h"


zyfjeff · 2018-08-01T15:13:21Z

@mattklein123 conn_manager_impl call sendLocalReply is head response, the body is empty, the other HTTP filter test conn_manager_impl didn't go, I don't know how to test, some HTTP filter integration testing, walked conn_manager_impl, I already inside the test case，The AsyncClient::sendLocalReply I didn't find a place to call

Can you give me some suggestions?

mattklein123 · 2018-08-01T21:09:03Z

The AsyncClient::sendLocalReply I didn't find a place to call

For AsyncClient, sendLocalReply() can be called in certain error cases where the router does a direct response. For example in this test (https://github.com/envoyproxy/envoy/blob/master/test/common/http/async_client_impl_test.cc#L466) I think if you change that to a HEAD request you will hit the code you modified.

@mattklein123 conn_manager_impl call sendLocalReply is head response, the body is empty, the other HTTP filter test conn_manager_impl didn't go, I don't know how to test, some HTTP filter integration testing, walked conn_manager_impl, I already inside the test case

It would be nice to have a conn_manager_impl_test for this, though nice job adding the integration tests, so I'm not going to block on it, but if you feel like it would be great. Check out https://github.com/envoyproxy/envoy/blob/master/test/common/http/conn_manager_impl_test.cc#L2880 and the tests around it. You should be able to setup a test like that, and do a local reply from within various filter callbacks which should drive the various flows.

Thank you!

zyfjeff · 2018-08-02T12:43:56Z

@mattklein123 AsyncClient and conn_manager_impl added some unittests

mattklein123

This is awesome, thanks. One small nit. @alyssawilk if you have a sec can you take a quick skim to see if you have any comments?

mattklein123 · 2018-08-02T17:03:00Z

test/common/http/utility_test.cc

+  Utility::sendLocalReply(false, callbacks, is_reset, Http::Code::PayloadTooLarge, "large", false);
+}
+
+TEST(HttpUtility, SendLocalReplayHeaderReply) {


typo "Replay". I think maybe you mean "SendLocalReplyHeadRequest" ?

yes, spelling mistakes

alyssawilk

Looks great overall. As always I'm a big fan of integration tests - I think we could msotly copy-paste IdleTimeoutIntegrationTest.PerStreamIdleTimeoutAfterDownstreamHeaders for and easy end to end regression test.

alyssawilk · 2018-08-02T17:13:27Z

source/common/http/utility.h

@@ -131,9 +131,10 @@ Http1Settings parseHttp1Settings(const envoy::api::v2::core::Http1ProtocolOption
 * @param response_code supplies the HTTP response code.
 * @param body_text supplies the optional body text which is sent using the text/plain content
 *                  type.
+ * @param is_head_request tells if this is a HEAD request


if this is a response to a HEAD request?

Yes, I fix it

mattklein123 · 2018-08-02T17:54:07Z

As always I'm a big fan of integration tests

There are 2 integration tests in the change (JWT filter and RBAC filter)

alyssawilk · 2018-08-02T17:59:28Z

Sorry, I could have said "we could also copy-paste" as I'd mildly prefer having an integration test which does not depend on filters. The one I suggested validates the behavior in the http connection manager for all 4 protocol combinations and permutations, which I believe has value.

mattklein123 · 2018-08-02T18:00:08Z

Ah yes, that would be good, agreed.

zyfjeff · 2018-08-06T11:29:07Z

@mattklein123 I met some problems when doing the integration test, has not yet been solved, can you help a look.

I started a Head request

  IntegrationStreamDecoderPtr setupPerStreamIdleTimeoutHeadRequestTest() {
    initialize();
    fake_upstreams_[0]->set_allow_unexpected_disconnects(true);
    codec_client_ = makeHttpConnection(makeClientConnection((lookupPort("http"))));
    auto encoder_decoder =
        codec_client_->startRequest(Http::TestHeaderMapImpl{{":method", "HEAD"},
                                                            {":path", "/test/long/url"},
                                                            {":scheme", "http"},
                                                            {":authority", "host"}});
    request_encoder_ = &encoder_decoder.first;
    auto response = std::move(encoder_decoder.second);
    fake_upstream_connection_ = fake_upstreams_[0]->waitForHttpConnection(*dispatcher_);
    upstream_request_ = fake_upstream_connection_->waitForNewStream(*dispatcher_);
    upstream_request_->waitForHeadersComplete();
    return response;
  }

Whether the head only request or other types of request end_stream is false

std::pair<Http::StreamEncoder&, IntegrationStreamDecoderPtr>
IntegrationCodecClient::startRequest(const Http::HeaderMap& headers) {
  auto response = std::make_unique<IntegrationStreamDecoder>(dispatcher_);
  Http::StreamEncoder& encoder = newStream(*response);
  encoder.getStream().addCallbacks(*response);
  // Here whether the head only request or with data, is false
  encoder.encodeHeaders(headers, false);
  flushWrite();
  return {encoder, std::move(response)};
}

This leads to the codec for HEAD request there is no set its response to the HEAD rsponse

void StreamEncoderImpl::encodeHeaders(const HeaderMap& headers, bool end_stream) {
.......

  if (end_stream) {
    endEncode();
  } else {
    connection_.flushOutput();
  }
}

void StreamEncoderImpl::endEncode() {
  if (chunk_encoding_) {
    connection_.buffer().add(LAST_CHUNK);
  }

  connection_.flushOutput();
  connection_.onEncodeComplete();
}

void ClientConnectionImpl::onEncodeComplete() {
  // Transfer head request state into the pending response before we reuse the encoder.
  pending_responses_.back().head_request_ = request_encoder_->headRequest();
}

The final test to find the response->compete() always returns false

[2018-08-06 11:23:15.466][16169][trace][http] source/common/http/http1/codec_impl.cc:341] [C48] parsing 0 bytes
[2018-08-06 11:23:15.466][16169][info][client] source/common/http/codec_client.cc:117] [C48] protocol error: http/1.1 protocol error: HPE_INVALID_EOF_STATE
[2018-08-06 11:23:15.466][16169][debug][client] source/common/http/codec_client.cc:81] [C48] disconnect. resetting 1 pending requests
[2018-08-06 11:23:15.466][16169][debug][client] source/common/http/codec_client.cc:104] [C48] request reset
[2018-08-06 11:23:15.466][16169][trace][main] source/common/event/dispatcher_impl.cc:126] item added to deferred deletion list (size=1)
[2018-08-06 11:23:15.466][16169][trace][main] source/common/event/dispatcher_impl.cc:52] clearing deferred deletion list (size=1)
test/integration/idle_timeout_integration_test.cc:104: Failure
Value of: response->complete()
  Actual: false
Expected: true

I repaired by changing the code

std::pair<Http::StreamEncoder&, IntegrationStreamDecoderPtr>
IntegrationCodecClient::startRequest(const Http::HeaderMap& headers) {
  auto response = std::make_unique<IntegrationStreamDecoder>(dispatcher_);
  Http::StreamEncoder& encoder = newStream(*response);
  encoder.getStream().addCallbacks(*response);
 // Determine whether it is a HEAD request to set the value of end_stream
  bool end_stream = (headers.Method()->value().c_str() == Http::Headers::get().MethodValues.Head);
  encoder.encodeHeaders(headers, end_stream);
  flushWrite();
  return {encoder, std::move(response)};
}

Test found that waitForDisconnect can not return

[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:341] [C48] parsing 130 bytes
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:439] [C48] message begin
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:309] [C48] completed header: key=content-length value=14
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:309] [C48] completed header: key=content-type value=text/plain
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:309] [C48] completed header: key=date value=Mon, 06 Aug 2018 11:28:09 GMT
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:405] [C48] headers complete
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:309] [C48] completed header: key=server value=envoy
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:426] [C48] message complete
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:729] [C48] message complete
[2018-08-06 11:28:09.849][22301][debug][client] source/common/http/codec_client.cc:94] [C48] response complete
[2018-08-06 11:28:09.849][22301][trace][main] source/common/event/dispatcher_impl.cc:126] item added to deferred deletion list (size=1)
[2018-08-06 11:28:09.849][22301][trace][http] source/common/http/http1/codec_impl.cc:362] [C48] parsed 130 bytes
[2018-08-06 11:28:09.849][22301][trace][main] source/common/event/dispatcher_impl.cc:52] clearing deferred deletion list (size=1)
[2018-08-06 11:28:09.849][22443][trace][connection] source/common/network/raw_buffer_socket.cc:67] [C49] write returns: 130
[2018-08-06 11:28:14.638][22438][debug][main] source/server/server.cc:119] flushing stats
[2018-08-06 11:28:19.639][22438][debug][main] source/server/server.cc:119] flushing stats
[2018-08-06 11:28:24.639][22438][debug][main] source/server/server.cc:119] flushing stats
[2018-08-06 11:28:29.639][22438][debug][main] source/server/server.cc:119] flushing stats
[2018-08-06 11:28:34.640][22438][debug][main] source/server/server.cc:119] flushing stats

alyssawilk · 2018-08-06T15:05:15Z

If you stick the whole test log with -l trace to pastebin I'll take a look. If there's no disconnect it's possible Envoy thinks it has handled the full request/response pair. Given startRequest I'd think it should be in "handling request" when the idle timeout comes and so end the connection. Given we (unfortunately) allow GET with body I assume we allow HEAD with body, but if I can see full logs I might be able to help you puzzle out what's going on.

zyfjeff · 2018-08-06T16:25:26Z

@alyssawilk test.log, Connection should be closed, I didn't see related connection through the netstat command information, through the log view, seems to be caused by connection was closed early.

I modify IntegrationCodecClient: : startRequest of encoder.EncodeHeaders (headers, false) To the encoder EncodeHeaders (headers, true) in latest master code, Also appear the same problem, log as follows
test.log.2

Modify the diff content is as follows:

diff --git a/test/integration/http_integration.cc b/test/integration/http_integration.cc
index 3f93b90..c756b49 100644
--- a/test/integration/http_integration.cc
+++ b/test/integration/http_integration.cc
@@ -137,7 +137,7 @@ IntegrationCodecClient::startRequest(const Http::HeaderMap& headers) {
   auto response = std::make_unique<IntegrationStreamDecoder>(dispatcher_);
   Http::StreamEncoder& encoder = newStream(*response);
   encoder.getStream().addCallbacks(*response);
-  encoder.encodeHeaders(headers, false);
+  encoder.encodeHeaders(headers, true);
   flushWrite();
   return {encoder, std::move(response)};
 }

Thank you very much can you help me to look at this problem

zyfjeff · 2018-08-09T06:14:52Z

@mattklein123 @alyssawilk I agree with your view, I re-adjusted the call point of connection_.onEncodeComplete() and put it in the place where the request header was received.

alyssawilk · 2018-08-09T14:54:15Z

test/integration/idle_timeout_integration_test.cc

@@ -43,6 +43,23 @@ class IdleTimeoutIntegrationTest : public HttpProtocolIntegrationTest {
    return response;
  }

+  IntegrationStreamDecoderPtr setupPerStreamIdleTimeoutHeadRequestTest() {


Also sorry, I should have thought of this before, but is this the same as setupPerStreamIdleTimeoutTest() apart from the method? If so, what do you think of changing setupPerStreamIdleTimeoutTest() -> setupPerStreamIdleTimeoutTest(cont char* method = "GET") to reduce similar code?

mattklein123 · 2018-08-10T00:43:39Z

source/common/http/http1/codec_impl.cc

+    // If this is an incomplete request (end_stream=false), then connection_.onEncodeComplete();
+    // will not be called in StreamEncoderImpl::encodeHeaders, head request state will not be passed
+    // to the corresponding response so here need to call this method
+    if (!end_stream)


nit: needs braces around this if statement.

I'm still not a fan of calling onEncodeComplete here as the encode is not necessarily complete. Can we just pass the state directly? This is what I was referring to previously. Can the the RequestStreamEncoderImpl hold a reference to the ClientConnection such that we can transfer this state directly? I think that might be cleaner?

You actually think that the name of onEncodeComplete will have some ambiguity. In fact, there is no complete encode.

Can I directly dynamic_cast from ConnectionImpl to ClientConnectionImpl and pass the state directly?

mattklein123 · 2018-08-10T04:09:17Z

source/common/http/http1/codec_impl.cc

+    // to the corresponding response so here need to call this method
+    if (!end_stream) {
+      auto c = dynamic_cast<ClientConnectionImpl*>(&connection_);
+      if (c) {


This should always be true. If we are going to do the dynamic cast we can just assume it is valid, though I'm not a huge fan of dynamic_cast for something like this. The other option is to store ClientConncetion in RequestStreamEncoderImpl (and server connection in ResponseStreamEncoderImpl and then have a base class virtual function connection() which returns the connection base class. @alyssawilk any opinions here?

I'd suggest that just like we have a virtual onEncodeComplete that the client/server connections implement, we could have a pure virtual onEncodeHeaders. We could handle passing on head status in the ClientConnectionImpl and the ServerConnectionImpl could just be a no-op for now.

I think that would be consistent with the style of the rest of the code base, and make it clear what part of the request/response pipeline is being handled.

alyssawilk

Looking good - almost there!

alyssawilk · 2018-08-14T13:15:19Z

source/common/http/http1/codec_impl.cc

-void ClientConnectionImpl::onEncodeComplete() {
-  // Transfer head request state into the pending response before we reuse the encoder.
-  pending_responses_.back().head_request_ = request_encoder_->headRequest();
+void ClientConnectionImpl::onEncodeComplete() {}


mild preference for moving this to the .h file

alyssawilk · 2018-08-14T13:15:51Z

source/common/http/http1/codec_impl.cc

-  pending_responses_.back().head_request_ = request_encoder_->headRequest();
+void ClientConnectionImpl::onEncodeComplete() {}
+
+void ClientConnectionImpl::onEncodeHeader(const HeaderMapImpl& headers) {


onEncodeHeaders for consistency with encodeHeaders?

Sorry, I think there was a misunderstanding on this one.
I was asking the name go from onEncodeHeader -> onEncodeHeaders (add an "s")
We don't need the same arguments, just Header with an S :-)
If you can remove the extra argument and change the name that would be great!

LGTM once this comment is addressed :-)

Sorry, I have never seen this comment (onEncodeHeader -> onEncodeHeaders (add an "s"), so I have not paid attention to this PR in recent days.

alyssawilk · 2018-08-14T13:16:12Z

source/common/http/http1/codec_impl.h

@@ -123,6 +122,11 @@ class ConnectionImpl : public virtual Connection, protected Logger::Loggable<Log
   */
  virtual void onEncodeComplete() PURE;

+  /**
+   * Called when header are start encoding.


Called when headers are encoded

alyssawilk · 2018-08-14T13:16:30Z

source/common/http/http1/codec_impl.h

@@ -303,6 +307,7 @@ class ServerConnectionImpl : public ServerConnection, public ConnectionImpl {

  // ConnectionImpl
  void onEncodeComplete() override;
+  void onEncodeHeader(const HeaderMapImpl& headers) override { UNREFERENCED_PARAMETER(headers); }


I think you can simplify this as
void onEncodeHeader(const HeaderMapImpl&) override { }

alyssawilk · 2018-08-14T13:17:12Z

source/extensions/filters/http/grpc_web/grpc_web_filter.cc

@@ -45,7 +45,6 @@ Http::FilterHeadersStatus GrpcWebFilter::decodeHeaders(Http::HeaderMap& headers,
    return Http::FilterHeadersStatus::Continue;
  }
  is_grpc_web_request_ = true;
-


I think you have a few whitespace removals you can revert

zyfjeff · 2018-08-14T15:17:59Z

@alyssawilk @mattklein123 Thank you for some review suggestions.

mattklein123

Generally looks great to me. Thank you! Will defer to @alyssawilk for further review/merge.

zyfjeff · 2018-08-20T03:16:22Z

@alyssawilk If this PR wants to merge, what else do I need to do?

alyssawilk

Cool, almost set. Please do one more master merge to pick up clang-6* and this should be good to go.

https://groups.google.com/forum/#!topic/envoy-dev/Y15Dd3MQ7Wc

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

* origin/master: (34 commits) fuzz: fixes oss-fuzz: 9895 (envoyproxy#4189) docs: added developer docs for pprof/tcmalloc testing (envoyproxy#4194) fix sometimes returns response body to HEAD requests (envoyproxy#3985) admin: fix config dump order (envoyproxy#4197) thrift: allow translation between downstream and upstream protocols (envoyproxy#4136) Use local_info.node() instead of bootstrap.node() whenever possible (envoyproxy#4120) upstream: allow extension protocol options to be used with http filters (envoyproxy#4165) [thrift_proxy] Replace local HeaderMap with Http::HeaderMap[Impl] (envoyproxy#4169) docs: improve gRPC-JSON filter doc (envoyproxy#4184) stats: refactoring MetricImpl without strings (envoyproxy#4190) fuzz: coverage profile generation instructions. (envoyproxy#4193) upstream: rebuild cluster when health check config is changed (envoyproxy#4075) build: use clang-6.0. (envoyproxy#4168) thrift_proxy: introduce header transport (envoyproxy#4082) tcp: allow connection pool callers to store protocol state (envoyproxy#4131) configs: match latest API changes (envoyproxy#4185) Fix wrong mock function name. (envoyproxy#4187) Bump yaml-cpp so it builds with Visual Studio 15.8 (envoyproxy#4182) Converting envoy configs to V2 (envoyproxy#2957) Add timestamp to HealthCheckEvent definition (envoyproxy#4119) ... Signed-off-by: Snow Pettersen <snowp@squareup.com>

zyfjeff requested a review from lizan as a code owner July 29, 2018 01:57

zyfjeff force-pushed the master branch from 7b7eacd to c130de3 Compare July 29, 2018 03:04

mattklein123 reviewed Jul 30, 2018

View reviewed changes

mattklein123 assigned alyssawilk and mattklein123 Jul 30, 2018

zyfjeff force-pushed the master branch 4 times, most recently from 78fa8e1 to 8e84cf8 Compare July 30, 2018 07:51

mattklein123 reviewed Jul 30, 2018

View reviewed changes

zyfjeff force-pushed the master branch 4 times, most recently from a7c8755 to 598212d Compare August 1, 2018 15:13

zyfjeff force-pushed the master branch from 598212d to f49df97 Compare August 1, 2018 15:17

zyfjeff force-pushed the master branch from f49df97 to 5049319 Compare August 2, 2018 12:24

zyfjeff force-pushed the master branch from 5049319 to d23fdca Compare August 2, 2018 15:00

mattklein123 reviewed Aug 2, 2018

View reviewed changes

alyssawilk reviewed Aug 2, 2018

View reviewed changes

zyfjeff force-pushed the master branch from d23fdca to cfdf3e0 Compare August 6, 2018 11:29

zyfjeff mentioned this pull request Aug 6, 2018

Envoy sometimes returns response body to HEAD requests #3932

Closed

zyfjeff force-pushed the master branch from 9856958 to daa10d7 Compare August 9, 2018 02:53

alyssawilk reviewed Aug 9, 2018

View reviewed changes

mattklein123 reviewed Aug 10, 2018

View reviewed changes

zyfjeff force-pushed the master branch from 23b1596 to b480d18 Compare August 10, 2018 02:10

mattklein123 reviewed Aug 10, 2018

View reviewed changes

alyssawilk requested changes Aug 14, 2018

View reviewed changes

zyfjeff force-pushed the master branch from e967ec2 to 758dcb6 Compare August 14, 2018 13:48

mattklein123 reviewed Aug 14, 2018

View reviewed changes

alyssawilk requested changes Aug 20, 2018

View reviewed changes

zyfjeff added 7 commits August 20, 2018 23:08

fix sometimes returns response body to HEAD requests

a1c22cf

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

Modify the call point of connection_.onEncodeComplete()

cf8bd87

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

Remove duplicated code

de67043

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

Add setHeadStateForLastResponse method

6cc889f

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

Add onEncodeHeader interface

5e8c918

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

Fix the bad code according to the review

183f14d

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

Change onEncodeHeader to onEncodeHeaders

a1f004e

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

zyfjeff requested a review from zuercher as a code owner August 20, 2018 15:36

zyfjeff force-pushed the master branch from b35e263 to a1f004e Compare August 20, 2018 15:40

alyssawilk previously approved these changes Aug 20, 2018

View reviewed changes

Fix build errors

b1a441c

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

zyfjeff dismissed alyssawilk’s stale review via b1a441c August 20, 2018 16:38

zyfjeff force-pushed the master branch from e21408b to b1a441c Compare August 20, 2018 16:38

Add method for fuzz test

2f5deb2

Signed-off-by: tianqian.zyf <tianqian.zyf@alibaba-inc.com>

alyssawilk approved these changes Aug 20, 2018

View reviewed changes

alyssawilk merged commit 0bdac76 into envoyproxy:master Aug 20, 2018

zyfjeff deleted the master branch July 22, 2019 08:49

		@@ -1,5 +1,6 @@
		#include "extensions/filters/http/jwt_authn/filter.h"

		#include "common/http/headers.h"

		@@ -1,5 +1,6 @@
		#include "extensions/filters/http/rbac/rbac_filter.h"

		#include "common/http/headers.h"

fix sometimes returns response body to HEAD requests #3985

fix sometimes returns response body to HEAD requests #3985

Conversation

zyfjeff commented Jul 29, 2018 • edited Loading

mattklein123 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

mattklein123 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zyfjeff Jul 31, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zyfjeff commented Aug 1, 2018 • edited Loading

mattklein123 commented Aug 1, 2018

zyfjeff commented Aug 2, 2018

mattklein123 left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

alyssawilk left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zyfjeff Aug 3, 2018 • edited Loading

Choose a reason for hiding this comment

mattklein123 commented Aug 2, 2018

alyssawilk commented Aug 2, 2018

mattklein123 commented Aug 2, 2018

zyfjeff commented Aug 6, 2018 • edited Loading

alyssawilk commented Aug 6, 2018

zyfjeff commented Aug 6, 2018

zyfjeff commented Aug 9, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zyfjeff Aug 10, 2018 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

alyssawilk left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

zyfjeff commented Aug 14, 2018

mattklein123 left a comment

Choose a reason for hiding this comment

zyfjeff commented Aug 20, 2018

alyssawilk left a comment

Choose a reason for hiding this comment

zyfjeff commented Jul 29, 2018 •

edited

Loading

zyfjeff Jul 31, 2018 •

edited

Loading

zyfjeff commented Aug 1, 2018 •

edited

Loading

zyfjeff Aug 3, 2018 •

edited

Loading

zyfjeff commented Aug 6, 2018 •

edited

Loading

zyfjeff commented Aug 9, 2018 •

edited

Loading

zyfjeff Aug 10, 2018 •

edited

Loading