Responses are constructed from requests using MakeResponse, MakeError, and
MakeErrorf. This ensures the response is always paired with the correct ID,
makes cases where there is no ID more explicit at the usage site, and
consolidates the handling of error introspection across transports.
The logic for unpacking errors and assigning JSON-RPC response types was
previously duplicated in three places. Consolidate it in the types package for
the RPC subsystem.
* update test cases
Instead of having the exported Request and Response type expose the version
directly, delegate version injection to an unexported shim. We already had the
shim; this just boosts it to the top level and does a bit more checking.
* Check JSON-RPC version marker.
The main change here is to use encoding/json to encode and decode RPC
parameters, rather than the custom tmjson package. This includes:
- Update the HTTP POST handler parameter handling.
- Add field tags to 64-bit integer types to get string encoding (to match amino/tmjson).
- Add marshalers to struct types that mention interfaces.
- Inject wrappers to decode interface arguments in RPC handlers.
* Rename rpctypes.Context to CallInfo.
Add methods to attach and recover this value from a context.Context.
* Rework RPC method handlers to accept "real" contexts.
- Replace *rpctypes.Context arguments with context.Context.
- Update usage of RPC context fields to use CallInfo.
* Update Caller interface and its documentation.
* Rework MapToRequest as ParamsToRequest.
The old interface returned the result as well as populating it. Nothing was
using this, so drop the duplicated value from the return signature. Clarify the
documentation on the Caller type.
Rework the MapToRequest helper to take an arbitrary value instead of only a
map. This is groundwork for getting rid of the custom marshaling code. For now,
however, the implementation preserves the existing behaviour for the map, until
we can replace those.
Apart from the tests for the websocket client, positional parameters are not
used by RPC clients. The server supports both arrays and objects, but the
client only needs to provide one or the other.
This continues the push of plumbing contexts through tendermint. I
attempted to find all goroutines in the production code (non-test) and
made sure that these threads would exit when their contexts were
canceled, and I believe this PR does that.
* docs: goleveldb is much more stable now
Refs https://github.com/syndtr/goleveldb/issues/226#issuecomment-682495490
* rpc/core/events: make sure WS client receives every event
previously, if the write buffer was full, the response would've been
lost without any trace (log msg, etc.)
* rpc/jsonrpc/server: set defaultWSWriteChanCapacity to 1
Refs #3905Closes#3829
setting write buffer capacity to 1 makes transactions count per block
more stable and also reduces the pauses length by 20s.
before: https://github.com/tendermint/tendermint/issues/3905#issuecomment-681854328 net.Read - 20s
after: net.Read - 0.66s
* rpc/jsonrpc/server: buffer writes and avoid io.ReadAll during read
Migrates the `rpc` package to use new JSON encoder in #4955. Branched off of that PR.
Tests pass, but I haven't done any manual testing beyond that. This should be handled as part of broader 0.34 testing.
Closes#4603
Commands used (VIM):
```
:args `rg -l errors.Wrap`
:argdo normal @q | update
```
where q is a macros rewriting the `errors.Wrap` to `fmt.Errorf`.
https://www.jsonrpc.org/specification
What is done in this PR:
JSONRPCClient: validate that Response.ID matches Request.ID I wanted
to do the same for the WSClient, but since we're sending events as
responses, not notifications, checking IDs would require storing
them in memory indefinitely (and we won't be able to remove them
upon client unsubscribing because ID is different then).
Request.ID is now optional. Notification is a Request without an ID.
Previously "" or 0 were considered as notifications
Remove #event suffix from ID from an event response (partially fixes
#2949) ID must be either string, int or null AND must be equal to
request's ID. Now, because we've implemented events as responses, WS
clients are tripping when they see Response.ID("0#event") !=
Request.ID("0"). Implementing events as requests would require a lot
of time (~ 2 days to completely rewrite WS client and server)
generate unique ID for each request
switch to integer IDs instead of "json-client-XYZ"
id=0 method=/subscribe
id=0 result=...
id=1 method=/abci_query
id=1 result=...
> send events (resulting from /subscribe) as requests+notifications (not
responses)
this will require a lot of work. probably not worth it
* rpc: generate an unique ID for each request
in conformance with JSON-RPC spec
* WSClient: check for unsolicited responses
* fix golangci warnings
* save commit
* fix errors
* remove ID from responses from subscribe
Refs #2949
* clients are safe for concurrent access
* tm-bench: switch to int ID
* fixes after my own review
* comment out sentIDs in WSClient
see commit body for the reason
* remove body.Close
it will be closed automatically
* stop ws connection outside of write/read routines
also, use t.Rate in tm-bench indexer when calculating ID
fix gocritic issues
* update swagger.yaml
* Apply suggestions from code review
* fix stylecheck and golint linter warnings
* update changelog
* update changelog2
* Make sure config.TimeoutBroadcastTxCommit < rpcserver.WriteTimeout()
* remove redundant comment
* libs/rpc/http_server: move Read/WriteTimeout into Config
* increase defaults for read/write timeouts
Based on this article
https://www.digitalocean.com/community/tutorials/how-to-optimize-nginx-configuration
* WriteTimeout should be larger than TimeoutBroadcastTxCommit
* set a deadline for subscribing to txs
* extract duration into const
* add two changelog entries
* Update CHANGELOG_PENDING.md
Co-Authored-By: melekes <anton.kalyaev@gmail.com>
* Update CHANGELOG_PENDING.md
Co-Authored-By: melekes <anton.kalyaev@gmail.com>
* 12 -> 10
* changelog
* changelog
* limit number of /subscribe clients and queries per client
Add the following config variables (under [rpc] section):
* max_subscription_clients
* max_subscriptions_per_client
* timeout_broadcast_tx_commit
Fixes#2826
new HTTPClient interface for subscriptions
finalize HTTPClient events interface
remove EventSubscriber
fix data race
```
WARNING: DATA RACE
Read at 0x00c000a36060 by goroutine 129:
github.com/tendermint/tendermint/rpc/client.(*Local).Subscribe.func1()
/go/src/github.com/tendermint/tendermint/rpc/client/localclient.go:168 +0x1f0
Previous write at 0x00c000a36060 by goroutine 132:
github.com/tendermint/tendermint/rpc/client.(*Local).Subscribe()
/go/src/github.com/tendermint/tendermint/rpc/client/localclient.go:191 +0x4e0
github.com/tendermint/tendermint/rpc/client.WaitForOneEvent()
/go/src/github.com/tendermint/tendermint/rpc/client/helpers.go:64 +0x178
github.com/tendermint/tendermint/rpc/client_test.TestTxEventsSentWithBroadcastTxSync.func1()
/go/src/github.com/tendermint/tendermint/rpc/client/event_test.go:139 +0x298
testing.tRunner()
/usr/local/go/src/testing/testing.go:827 +0x162
Goroutine 129 (running) created at:
github.com/tendermint/tendermint/rpc/client.(*Local).Subscribe()
/go/src/github.com/tendermint/tendermint/rpc/client/localclient.go:164 +0x4b7
github.com/tendermint/tendermint/rpc/client.WaitForOneEvent()
/go/src/github.com/tendermint/tendermint/rpc/client/helpers.go:64 +0x178
github.com/tendermint/tendermint/rpc/client_test.TestTxEventsSentWithBroadcastTxSync.func1()
/go/src/github.com/tendermint/tendermint/rpc/client/event_test.go:139 +0x298
testing.tRunner()
/usr/local/go/src/testing/testing.go:827 +0x162
Goroutine 132 (running) created at:
testing.(*T).Run()
/usr/local/go/src/testing/testing.go:878 +0x659
github.com/tendermint/tendermint/rpc/client_test.TestTxEventsSentWithBroadcastTxSync()
/go/src/github.com/tendermint/tendermint/rpc/client/event_test.go:119 +0x186
testing.tRunner()
/usr/local/go/src/testing/testing.go:827 +0x162
==================
```
lite client works (tested manually)
godoc comments
httpclient: do not close the out channel
use TimeoutBroadcastTxCommit
no timeout for unsubscribe
but 1s Local (5s HTTP) timeout for resubscribe
format code
change Subscribe#out cap to 1
and replace config vars with RPCConfig
TimeoutBroadcastTxCommit can't be greater than rpcserver.WriteTimeout
rpc: Context as first parameter to all functions
reformat code
fixes after my own review
fixes after Ethan's review
add test stubs
fix config.toml
* fixes after manual testing
- rpc: do not recommend to use BroadcastTxCommit because it's slow and wastes
Tendermint resources (pubsub)
- rpc: better error in Subscribe and BroadcastTxCommit
- HTTPClient: do not resubscribe if err = ErrAlreadySubscribed
* fixes after Ismail's review
* Update rpc/grpc/grpc_test.go
Co-Authored-By: melekes <anton.kalyaev@gmail.com>
* green pubsub tests :OK:
* get rid of clientToQueryMap
* Subscribe and SubscribeUnbuffered
* start adapting other pkgs to new pubsub
* nope
* rename MsgAndTags to Message
* remove TagMap
it does not bring any additional benefits
* bring back EventSubscriber
* fix test
* fix data race in TestStartNextHeightCorrectly
```
Write at 0x00c0001c7418 by goroutine 796:
github.com/tendermint/tendermint/consensus.TestStartNextHeightCorrectly()
/go/src/github.com/tendermint/tendermint/consensus/state_test.go:1296 +0xad
testing.tRunner()
/usr/local/go/src/testing/testing.go:827 +0x162
Previous read at 0x00c0001c7418 by goroutine 858:
github.com/tendermint/tendermint/consensus.(*ConsensusState).addVote()
/go/src/github.com/tendermint/tendermint/consensus/state.go:1631 +0x1366
github.com/tendermint/tendermint/consensus.(*ConsensusState).tryAddVote()
/go/src/github.com/tendermint/tendermint/consensus/state.go:1476 +0x8f
github.com/tendermint/tendermint/consensus.(*ConsensusState).handleMsg()
/go/src/github.com/tendermint/tendermint/consensus/state.go:667 +0xa1e
github.com/tendermint/tendermint/consensus.(*ConsensusState).receiveRoutine()
/go/src/github.com/tendermint/tendermint/consensus/state.go:628 +0x794
Goroutine 796 (running) created at:
testing.(*T).Run()
/usr/local/go/src/testing/testing.go:878 +0x659
testing.runTests.func1()
/usr/local/go/src/testing/testing.go:1119 +0xa8
testing.tRunner()
/usr/local/go/src/testing/testing.go:827 +0x162
testing.runTests()
/usr/local/go/src/testing/testing.go:1117 +0x4ee
testing.(*M).Run()
/usr/local/go/src/testing/testing.go:1034 +0x2ee
main.main()
_testmain.go:214 +0x332
Goroutine 858 (running) created at:
github.com/tendermint/tendermint/consensus.(*ConsensusState).startRoutines()
/go/src/github.com/tendermint/tendermint/consensus/state.go:334 +0x221
github.com/tendermint/tendermint/consensus.startTestRound()
/go/src/github.com/tendermint/tendermint/consensus/common_test.go:122 +0x63
github.com/tendermint/tendermint/consensus.TestStateFullRound1()
/go/src/github.com/tendermint/tendermint/consensus/state_test.go:255 +0x397
testing.tRunner()
/usr/local/go/src/testing/testing.go:827 +0x162
```
* fixes after my own review
* fix formatting
* wait 100ms before kicking a subscriber out
+ a test for indexer_service
* fixes after my second review
* no timeout
* add changelog entries
* fix merge conflicts
* fix typos after Thane's review
Co-Authored-By: melekes <anton.kalyaev@gmail.com>
* reformat code
* rewrite indexer service in the attempt to fix failing test
https://github.com/tendermint/tendermint/pull/3227/#issuecomment-462316527
* Revert "rewrite indexer service in the attempt to fix failing test"
This reverts commit 0d9107a098.
* another attempt to fix indexer
* fixes after Ethan's review
* use unbuffered channel when indexing transactions
Refs https://github.com/tendermint/tendermint/pull/3227#discussion_r258786716
* add a comment for EventBus#SubscribeUnbuffered
* format code
* Fixed accepting integer IDs in requests for Tendermint RPC server (#2366)
* added a wrapper interface `jsonrpcid` that represents both string and int IDs in JSON-RPC requests/responses + custom JSON unmarshallers
* changed client-side code in RPC that uses it
* added extra tests for integer IDs
* updated CHANGELOG_PENDING, as suggested by PR instructions
* addressed PR comments
* added table driven tests for request type marshalling/unmarshalling
* expanded handler test to check IDs
* changed pending changelog note
* changed json rpc request/response unmarshalling to use empty interfaces and type switches on ID
* some cleanup
Refs #951
Jae: I don't know a good way to catch these errors in general, but
forcing pubsub's internal channel to have a capacity of 0 will reveal
bugs sooner, if the subscriber also has a 0 or small capacity ch to pull
from.
* use increment and decrement operators.
* remove unnecessary else branches.
* fix package comment with leading space.
* fix receiver names.
* fix error strings.
* remove omittable code.
* remove redundant return statement.
* Revert changes (code is generated.)
* use cfg as receiver name for all config-related types.
* use lsi as the receiver name for the LastSignedInfo type.
comment out failing consensus tests for now
rewrite rpc httpclient to use new pubsub package
import pubsub as tmpubsub, query as tmquery
make event IDs constants
EventKey -> EventTypeKey
rename EventsPubsub to PubSub
mempool does not use pubsub
rename eventsSub to pubsub
new subscribe API
fix channel size issues and consensus tests bugs
refactor rpc client
add missing discardFromChan method
add mutex
rename pubsub to eventBus
remove IsRunning from WSRPCConnection interface (not needed)
add a comment in broadcastNewRoundStepsAndVotes
rename registerEventCallbacks to broadcastNewRoundStepsAndVotes
See https://dave.cheney.net/2014/03/19/channel-axioms
stop eventBuses after reactor tests
remove unnecessary Unsubscribe
return subscribe helper function
move discardFromChan to where it is used
subscribe now returns an err
this gives us ability to refuse to subscribe if pubsub is at its max
capacity.
use context for control overflow
cache queries
handle err when subscribing in replay_test
rename testClientID to testSubscriber
extract var
set channel buffer capacity to 1 in replay_file
fix byzantine_test
unsubscribe from single event, not all events
refactor httpclient to return events to appropriate channels
return failing testReplayCrashBeforeWriteVote test
fix TestValidatorSetChanges
refactor code a bit
fix testReplayCrashBeforeWriteVote
add comment
fix TestValidatorSetChanges
fixes from Bucky's review
update comment [ci skip]
test TxEventBuffer
update changelog
fix TestValidatorSetChanges (2nd attempt)
only do wg.Done when no errors
benchmark event bus
create pubsub server inside NewEventBus
only expose config params (later if needed)
set buffer capacity to 0 so we are not testing cache
new tx event format: key = "Tx" plus a tag {"tx.hash": XYZ}
This should allow to subscribe to all transactions! or a specific one
using a query: "tm.events.type = Tx and tx.hash = '013ABF99434...'"
use TimeoutCommit instead of afterPublishEventNewBlockTimeout
TimeoutCommit is the time a node waits after committing a block, before
it goes into the next height. So it will finish everything from the last
block, but then wait a bit. The idea is this gives it time to hear more
votes from other validators, to strengthen the commit it includes in the
next block. But it also gives it time to hear about new transactions.
waitForBlockWithUpdatedVals
rewrite WAL crash tests
Task:
test that we can recover from any WAL crash.
Solution:
the old tests were relying on event hub being run in the same thread (we
were injecting the private validator's last signature).
when considering a rewrite, we considered two possible solutions: write
a "fuzzy" testing system where WAL is crashing upon receiving a new
message, or inject failures and trigger them in tests using something
like https://github.com/coreos/gofail.
remove sleep
no cs.Lock around wal.Save
test different cases (empty block, non-empty block, ...)
comments
add comments
test 4 cases: empty block, non-empty block, non-empty block with smaller part size, many blocks
fixes as per Bucky's last review
reset subscriptions on UnsubscribeAll
use a simple counter to track message for which we panicked
also, set a smaller part size for all test cases
from the spec:
This member is REQUIRED.
It MUST be the same as the value of the id member in the Request Object.
If there was an error in detecting the id in the Request object (e.g. Parse error/Invalid Request), it MUST be Null.
```
data
A Primitive or Structured value that contains additional information about the error.
This may be omitted.
The value of this member is defined by the Server (e.g. detailed error information, nested errors etc.).
```
More changes:
- remove Client interface (reason: empty)
- introduce HTTPClient interface, which can be used for both ClientURI
and ClientJSONRPC clients (so our users don't have to create their own) (Refs #8)
- rename integration tests script to `integration_test.sh`
- do not update deps on `get_deps`
M. J. Fromberger
Sam Kleinman
Anton Kaliaev
Bipul Prasad
Callum Waters
Erik Grinaker
Ismail Khoffi
Tomas Tauber
Zach Ramsay
Liamsi
Jae Kwon
Thomas Corbière
Ethan Buchman
Silas Davis
Dave Bryson
Ethan Frey