Commit Graph

121 Commits

Author SHA1 Message Date
Jean de Klerk a402911c6f
internal: resetTransport connect deadline is across addresses (#2540)
internal: resetTransport connect deadline is across addresses

Currently, the connect deadline is recalculated per-address. This PR amends
that behavior such that all addresses for a single connection attempt share
the same deadline.

Fixes #2462
2019-02-11 17:12:42 -07:00
Menghan Li ec9c18c8c6
internal: split StateRecordingBalancer in test to balancer and builder (#2578)
And instead of setting state notify channel in balancer, create a new notify
channel at Build.

fixes #2576
2019-01-18 10:21:46 -08:00
Jean de Klerk 76cc50721c
internal: rewrite TestDialWithMultipleBackendsNotSendingServerPreface (#2438)
- Remove the slice of servers approach, since there's specific
logic at server 2 that's different from server 1. This has the
advantage of making the test more readable without sacrificing
anything (given the previous point).

- Defer server close at initialization time instead of at the
end.

- Remove a time.Sleep(time.Second): use timeout + select around
serverDone instead.

- Use a goroutine to keep the connection reading, instead of
using a for loop in the server goroutine. This causes the
defer close(server2Done) to happen immediately after preface
is sent, which combined with the aforementioned time.Sleep
removal causes the test to go from 1.00s to ~0.05s.
2019-01-08 16:48:09 -08:00
Doug Fawley 0a391ff2b7
grpctest: add new package to manage tests and support per-test setup/teardown (#2523)
- Migrate `grpc` & `grpc/test` packages to use `Teardown` support to guarantee `leakcheck` is used
2019-01-07 14:24:56 -08:00
Jean de Klerk 1b41b79fd1
internal: refactor transport to single retry mechanism (#2461)
Previously, the transport was able to reset via the retry loop,
or via the event closures calling resetTransport. This meant
a very large amount of synchronization was necessary: one
reset meant the other had to not reset; state had to be kept
at the addrconn; and very subtle interactions were hard to
reason about.

This change removes the ability for event closures to directly
reset the transport. Instead, they signal to to the retry
loop about the event, and the retry loop is always the single
place that retries occur.

This also allows us to refactor the address switching logic
into a much simpler for loop inside the retry loop instead of
using addrConn state to keep track of an index.
2018-12-17 13:10:13 -08:00
Doug Fawley f3eb5bc06e
client: add GRPC_GO_REQUIRE_HANDSHAKE options to control connection behavior (#2464)
Possible settings of this environment variable:

- "hybrid" (default; removed after the 1.17 release): do not wait for handshake before considering a connection ready, but wait before considering successful.

- "on" (default after the 1.17 release): wait for handshake before considering a connection ready/successful.

- "off": do not wait for handshake before considering a connection ready/successful.

This setting will be completely removed after the 1.18 release, and "on" will be the only supported behavior.
2018-11-26 15:06:46 -08:00
Doug Fawley 04ea82009c
cleanup: replace "x/net/context" import with "context" (#2439) 2018-11-12 13:30:41 -08:00
Jean de Klerk 04c0c4d299
internal: fix client send preface problems (#2380)
internal: fix client send preface problems

This CL fixes three problems:

- In clientconn_state_transitions_test.go, sometimes tests would flake because there's not enough buffer to send client side settings, causing the connection to unpredictably enter TRANSIENT FAILURE. Each time we set up a server to send SETTINGS, we should also set up the server to read. This allows the client to successfully send its SETTINGS, unflaking the test.

- In clientconn.go, we incorrectly transitioned into TRANSIENT FAILURE when creating an http2client returned an error. This should be handled in the outer resetTransport main reset loop. The reason this became a problem is that the outer resetTransport has very specific conditions around when to transition into TRANSIENT FAILURE that the egregious transition did not have. So, it could transition into TRANSIENT FAILURE after failing to dial, even if it was trying to connect to a non-final address in the list of addresses.

- In clientconn.go, we incorrectly stay in CONNECTING after `createTransport` when a server sends its connection preface but the client is not able to send its connection preface. This CL causes the addrconn to correctly enter TRANSIENT FAILURE when `createTransport` fails, even if a server preface was received. It does so by making ac.successfulHandshake to consider both server preface received as well as client preface sent.
2018-10-18 14:31:34 -07:00
Jean de Klerk 1ca9df53a7
internal: clean up and unflake state transitions test (#2366)
internal: clean up and unflake state transitions test

Switches state transitions test to using a notification from a custom load
balancer, instead of relying on waiting for laggy balancer state updates.

Also generally adds more coverage around state transitions and a framework
for easily adding more of these kinds of tests.

Fixes #2348
2018-10-12 15:22:27 -07:00
Menghan Li cb11627444
clientconn: not panic when service config updated while closing (#2371)
Closing `ClientConn` sets `balancerWrapper` to nil.
If service config switches balancer, the new balancer will be notified of the existing addresses.

When these two happens together, there's a chance that a method will be called on the nil `balancerWrapper`. This change adds a check to make sure that never happens. 

fixes #2367
2018-10-11 11:43:16 -07:00
Jean de Klerk 8d75951f9b
Fix onclose state transition (#2346)
internal: fix onClose state transitions

When onClose occurs during WaitForHandshake, it should immediately
exit createTransport instead of leaving CONNECTING and entering READY.

Furthermore, when any onClose happens, the state should change to
TRANSIENT FAILURE.

Fixes #2340
Fixes #2341

Also fixes an unreported bug in which entering READY causes a
Dial call to end prematurely, instead of blocking until a READY
transport is found.
2018-10-08 16:58:54 -06:00
dfawley cc41663c52
client: fix panic caused by WithWaitForHandshake and related races (#2336) 2018-09-27 14:23:29 -07:00
Jean de Klerk 82f263dc2f internal: fix race between cancel and shutdown
This fixes a race in ac.tearDown and ac.resetTransport. If ac.resetTransport
is in backoff when ac.tearDown occurs, there's a race between the state
changing to Shutdown and ac.resetTransport calling ac.createTransport.

This fixes it by returning when ac.resetTransport encounters an error
during ac.nextAddr (specifically ac.ctx.Error()). It also fixes it by
making sure that ac.tearDown changes state to Shutdown before canceling
the context.

Both fixes were implemented because they both seem to be valuable
standalone additions: the former makes ac.resetTransport more
understandable and less dependent on behavior happening elsewhere,
and the latter makes ac.tearDown more correct.

Finally, TestDialParseTargetUnknownScheme had its buffer removed; the
buffer was likely added a while ago to assuage this issue. It should
not be necessary anymore.
2018-09-24 10:46:24 -07:00
Jean de Klerk 0baa06717e
tests: fix goroutine leak (#2317)
tests: fix goroutine leak

If TestResetConnectBackoff fails, the resetTransport goroutine will be
stuck dialing and subsequently the goroutine will be leaked. This is
all despite the test including `defer cc.Close()`:

- defer cc.Close() will cause ac.cancel to be called
- ac.context will be appropriately cancelled
- ac.context is correctly the context that gets passed to the dialer
- However, the WithDialer throws away the context and only passes its
deadline, which is for `backoffForever{}` is math.MaxInt64. So, even
though teardown occurs, the resetTransport goroutine will still be
stuck dialing.

This CL adds a small amendment: before performing leakcheck, attempt
to take an item off the synchronous `dials` channel. Either the tests
passed and there is no item, or the tests failed and there is one.
2018-09-20 15:26:05 -07:00
dfawley d2aec4d7de
client: Add ClientConn.ResetConnectBackoff to force reconnections on demand (#2273)
Fixes #1037
2018-08-27 13:21:48 -07:00
Dustin Spicuzza 91c7ef84b5 client: fix FailOnNonTempDialError and add a test for it (#2276) 2018-08-27 10:28:41 -07:00
lyuxuan 980d9e0348
ClientConn: add Target() returning target string (#2233) 2018-07-23 16:19:11 -07:00
mmukhi cedd9131e8
Fix test: wait on server to signal successful accept. (#2183) 2018-06-28 11:19:52 -07:00
Menghan Li 24f3cca1ff
internal: move backoff to internal (#2141)
So other components such as grpclb can reuse the backoff implementation.
2018-06-13 16:07:37 -07:00
Jean de Klerk 0e5a36b652
internal: move leakcheck to internal/ (#2129)
internal: move leakcheck to internal/
2018-06-07 16:57:56 -07:00
lyuxuan 854695bef0
client: introduce WithDisableServiceConfig DialOption (#2010) 2018-05-08 10:28:26 -07:00
Jean de Klerk e09cbb70fe
internal: better test names (#2043) 2018-05-03 13:17:45 -07:00
mmukhi 031ee13cfe
Fix Test: Update the deadline since small deadlines are prone to flakes on Travis. (#1932) 2018-03-20 16:46:10 -07:00
mmukhi 207e2760fd
Fix test race: Atomically access minConnecTimout in testing environment. (#1897) 2018-03-07 10:36:17 -08:00
mmukhi 7c5299d71e
Fix flaky test: TestCloseConnectionWhenServerPrefaceNotReceived (#1870)
* Atomically update minConnectTimeout in test.

* Refactor the flaky test.

* Post review update

* mend
2018-03-02 13:39:55 -08:00
Menghan Li 37346e3181
Revert "Add WithResolverUserOptions for custom resolver build options" (#1839)
This reverts commit ff1be3fcc5.
2018-02-05 12:52:35 -08:00
mmukhi 65c901e458
Fix flakey test. (#1771) 2017-12-28 09:31:16 -08:00
mmukhi 6610f9a340
Fix test: Data race while resetting global var. (#1748) 2017-12-18 15:27:23 -08:00
mmukhi b8cf13ea06
Make sure all goroutines have ended before restoring global vars. (#1732) 2017-12-14 13:23:05 -08:00
Menghan Li ff1be3fcc5
Add WithResolverUserOptions for custom resolver build options (#1711) 2017-12-06 15:54:01 -08:00
mmukhi ddbb27e545
client: backoff before reconnecting if an HTTP2 server preface was not received (#1648) 2017-12-01 09:55:42 -08:00
Menghan Li 10873b30bf
Fix panics on balancer and resolver updates (#1684)
- NewAddress with empty list (addrConn with an empty address list)
 - NewServiceConfig to switch balancer before the first balancer is built
2017-11-22 13:59:20 -08:00
Menghan Li a353537ff5 Register and use default balancers and resolvers (#1551) 2017-10-19 11:32:06 -07:00
Menghan Li d46a3655c4 Add leak goroutine checking to grpc/balancer tests (#1497) 2017-09-07 14:30:05 -07:00
Menghan Li 8233e124e4 Add new Resolver and Balancer APIs (gRFC L9) (#1408)
- Add package balancer and resolver.
 - Change ClientConn internals to new APIs and adds a wrapper for v1 balancer.
2017-08-31 10:59:09 -07:00
dfawley e60698345e Fix context warnings from govet. (#1486)
Pre-req work for #1484
2017-08-29 11:04:15 -07:00
Menghan Li e81b5698fd Add and use connectivity package for states (#1430)
* Add and use connectivity package
* Mark cc state APIs as experimental
2017-08-09 10:31:12 -07:00
Andrew Lytvynov 4e56696c6c Fix a goroutine leak in DialContext (#1424)
A leak happens when DialContext times out before a balancer returns any
addresses or before a successful connection is established.

The loop in ClientConn.lbWatcher breaks and doneChan never gets closed.
2017-08-04 13:40:50 -07:00
Menghan Li b463cc3276 Call cancel on contexts in tests (#1412) 2017-08-02 10:43:35 -07:00
田欧 ca9e0c3458 Add testdata package and unify testdata to only one dir (#1297) 2017-07-25 10:24:45 -07:00
mmukhi a5d184a8a1 Expose ConnectivityState of a ClientConn. (#1385) 2017-07-24 15:00:53 -07:00
Jan Tattermusch ddbf6c46a6 autofix license notice 2017-06-08 14:42:19 +02:00
Tamir Duberstein 3773797869 Travis: add staticcheck (#1019)
Also only run golint and go vet in Go 1.8, and fix some vet failures.
2017-05-15 17:05:27 -07:00
MakMukhi bfa5dd27dc Client should update keepalive parameters upon receiving GoAway with … (#1169)
* Client should update keepalive parameters upon receiving GoAway with EnhanceYourCalm and debug data of too_many_pings.
2017-04-10 14:33:51 -07:00
Menghan Li 0df08a7a03 :authority should include port number (#1123) 2017-03-28 11:09:23 -07:00
dfawley c5a5dbc500 Don't return an error from dial if the balancer returns no initial servers (#1112)
This modifies the WithBlock behavior somewhat to block until there is at least
one valid connection.  Previously, each connection would be made serially until
all had completed successfully, with any errors returned to the caller.  Errors
are now only returned due to connecting to a backend if a balancer is not used,
or if there is an error starting the balancer itself.

Fixes #976
2017-03-21 11:35:53 -07:00
Chris Roche 4ad16bc34a Authority overwrite only works if TLS is not present 2017-02-13 10:25:06 -08:00
Chris Roche c7430a063e Only override :authority for insecure dials 2017-02-09 10:46:00 -08:00
Chris Roche 84bee50bda Add DialOption to overwrite :authority pseudo-header
The :authority pseudo-header for a gRPC Client defaults to the host
portion of the dialed target and can only be overwritten by providing a
TransportCredentials. However, there are cases where setting this header
independent of any tranport security is valid. In my particular case,
in order to leverage Envoy for request routing, the cluster/service name
must be provided in the :authority header. This may also be useful in a
testing context.

This patch adds a DialOption to overwrite the authority header,
even if TransportCredentials are provided (I'd imagine you'd only ever
need to specify one or the other).
2017-02-03 17:29:18 -08:00
Qi Zhao 09aecb094e Add the initial service config support (#1009)
* Add the initial service config support

* start scWatcher later

* remove timeoutCh

* address the comments

* deal with dial timeout

* defer cancel for the newly created context for correct lifetime management

* fix the defer order

* added other 2 missing cancels
2016-12-19 16:31:00 -08:00
Menghan Li 4f404a1c98 fix a circular dependency in clientconn_test 2016-11-18 15:01:12 -08:00
Menghan Li c31bccc236 add FailOnNonTempDialError to control if gRPC should fail on non-temp dial error 2016-11-15 11:14:07 -08:00
Menghan Li df5d48b597 fix comments 2016-11-09 17:35:47 -08:00
Menghan Li 947e436ef4 return non temporary connection error if dialer returns non temprary errors 2016-11-09 17:25:46 -08:00
iamqizhao f02984b7c6 Basic support of grpclb 2016-09-19 15:11:57 -07:00
Menghan Li 979f41603b review fix 2016-09-12 10:35:11 -07:00
Menghan Li 726462e0cc fix TestTLSServerNameOverwrite timeout error 2016-09-08 17:26:58 -07:00
Menghan Li 74f7afb1f9 Move balancer initialization into a goroutine 2016-09-08 15:13:50 -07:00
Menghan Li a00cbfeab5 Overwrite authority if creds servername is specified 2016-09-06 11:23:30 -07:00
iamqizhao 7873a050d4 fix the test 2016-08-24 10:46:06 -07:00
iamqizhao 7eae19acb7 fix the issue 2016-08-23 19:23:04 -07:00
Spencer Kimball b38541aeb0 Implement DialContext to afford caller option of managing cancelation 2016-08-16 16:57:59 -04:00
iamqizhao 6205cb25ab fix some bugs 2016-07-13 18:05:02 -07:00
Menghan Li d9c8fb446d Change errCredentialsMisuse to errCredentialsConflict and errTransportCredentialsMissing 2016-06-08 13:53:41 -07:00
Menghan Li 6404c49192 Make TransportAuthenticator not embed Credentials 2016-06-06 16:24:46 -07:00
iamqizhao 404e9b67de dial time out 2016-06-06 12:08:11 -07:00
iamqizhao 9dc3da0633 make downErr for Balancer down closure 2016-05-25 11:28:45 -07:00
iamqizhao 9cec19a4d4 make setDefault(...) non-member function 2016-04-18 13:15:27 -07:00
Stephen J Day 8ef1dcabab backoff: make DefaultBackoffConfig a concrete value
To enforce immutability of the `DefaultBackoffConfig`, we've made it a
concrete value. While fields can still be set directly on the value,
taking a copy will not incidentally pull a reference to the variable.

Signed-off-by: Stephen J Day <stephen.day@docker.com>
2016-04-18 11:33:39 -07:00
Stephen J Day 9ff38e9093 backoff: set default values on BackoffConfig
Because most of the fields on `BackoffConfig` are unexported, correctly
using the config requires copying from the default. This sets the
defaults appropriately and falls back to a default if MaxDelay is
negative or zero.

Tests are added to ensure that the backoff is set correctly in common
use cases.

Signedroff-by: Stephen J Day <stephen.day@docker.com>
Signed-off-by: Stephen J Day <stephen.day@docker.com>
2016-04-15 16:12:47 -07:00
iamqizhao cb8841c36e move some tests from end2end.go to clientconn_test.go 2015-11-06 11:33:13 -08:00