go-ethereum/rpc/client_test.go

984 lines
26 KiB
Go
Raw Normal View History

// Copyright 2016 The go-ethereum Authors
// This file is part of the go-ethereum library.
//
// The go-ethereum library is free software: you can redistribute it and/or modify
// it under the terms of the GNU Lesser General Public License as published by
// the Free Software Foundation, either version 3 of the License, or
// (at your option) any later version.
//
// The go-ethereum library is distributed in the hope that it will be useful,
// but WITHOUT ANY WARRANTY; without even the implied warranty of
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
// GNU Lesser General Public License for more details.
//
// You should have received a copy of the GNU Lesser General Public License
// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
package rpc
import (
"context"
"encoding/json"
"errors"
"fmt"
"math/rand"
"net"
"net/http"
"net/http/httptest"
"os"
"reflect"
"runtime"
"strings"
"sync"
"testing"
"time"
"github.com/davecgh/go-spew/spew"
"github.com/ethereum/go-ethereum/log"
)
func TestClientRequest(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
var resp echoResult
if err := client.Call(&resp, "test_echo", "hello", 10, &echoArgs{"world"}); err != nil {
t.Fatal(err)
}
if !reflect.DeepEqual(resp, echoResult{"hello", 10, &echoArgs{"world"}}) {
t.Errorf("incorrect result %#v", resp)
}
}
func TestClientResponseType(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
if err := client.Call(nil, "test_echo", "hello", 10, &echoArgs{"world"}); err != nil {
t.Errorf("Passing nil as result should be fine, but got an error: %v", err)
}
var resultVar echoResult
// Note: passing the var, not a ref
err := client.Call(resultVar, "test_echo", "hello", 10, &echoArgs{"world"})
if err == nil {
t.Error("Passing a var as result should be an error")
}
}
// This test checks calling a method that returns 'null'.
func TestClientNullResponse(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
var result json.RawMessage
if err := client.Call(&result, "test_null"); err != nil {
t.Fatal(err)
}
if result == nil {
t.Fatal("Expected non-nil result")
}
if !reflect.DeepEqual(result, json.RawMessage("null")) {
t.Errorf("Expected null, got %s", result)
}
}
internal/ethapi: return revert reason for eth_call (#21083) * internal/ethapi: return revert reason for eth_call * internal/ethapi: moved revert reason logic to doCall * accounts/abi/bind/backends: added revert reason logic to simulated backend * internal/ethapi: fixed linting error * internal/ethapi: check if require reason can be unpacked * internal/ethapi: better error logic * internal/ethapi: simplify logic * internal/ethapi: return vmError() * internal/ethapi: move handling of revert out of docall * graphql: removed revert logic until spec change * rpc: internal/ethapi: added custom error types * graphql: use returndata instead of return Return() checks if there is an error. If an error is found, we return nil. For most use cases it can be beneficial to return the output even if there was an error. This code should be changed anyway once the spec supports error reasons in graphql responses * accounts/abi/bind/backends: added tests for revert reason * internal/ethapi: add errorCode to revert error * internal/ethapi: add errorCode of 3 to revertError * internal/ethapi: unified estimateGasErrors, simplified logic * internal/ethapi: unified handling of errors in DoEstimateGas * rpc: print error data field * accounts/abi/bind/backends: unify simulatedBackend and RPC * internal/ethapi: added binary data to revertError data * internal/ethapi: refactored unpacking logic into newRevertError * accounts/abi/bind/backends: fix EstimateGas * accounts, console, internal, rpc: minor error interface cleanups * Revert "accounts, console, internal, rpc: minor error interface cleanups" This reverts commit 2d3ef53c5304e429a04983210a417c1f4e0dafb7. * re-apply the good parts of 2d3ef53c53 * rpc: add test for returning server error data from client Co-authored-by: rjl493456442 <garyrong0905@gmail.com> Co-authored-by: Péter Szilágyi <peterke@gmail.com> Co-authored-by: Felix Lange <fjl@twurst.com>
2020-06-08 03:09:49 -05:00
// This test checks that server-returned errors with code and data come out of Client.Call.
func TestClientErrorData(t *testing.T) {
t.Parallel()
internal/ethapi: return revert reason for eth_call (#21083) * internal/ethapi: return revert reason for eth_call * internal/ethapi: moved revert reason logic to doCall * accounts/abi/bind/backends: added revert reason logic to simulated backend * internal/ethapi: fixed linting error * internal/ethapi: check if require reason can be unpacked * internal/ethapi: better error logic * internal/ethapi: simplify logic * internal/ethapi: return vmError() * internal/ethapi: move handling of revert out of docall * graphql: removed revert logic until spec change * rpc: internal/ethapi: added custom error types * graphql: use returndata instead of return Return() checks if there is an error. If an error is found, we return nil. For most use cases it can be beneficial to return the output even if there was an error. This code should be changed anyway once the spec supports error reasons in graphql responses * accounts/abi/bind/backends: added tests for revert reason * internal/ethapi: add errorCode to revert error * internal/ethapi: add errorCode of 3 to revertError * internal/ethapi: unified estimateGasErrors, simplified logic * internal/ethapi: unified handling of errors in DoEstimateGas * rpc: print error data field * accounts/abi/bind/backends: unify simulatedBackend and RPC * internal/ethapi: added binary data to revertError data * internal/ethapi: refactored unpacking logic into newRevertError * accounts/abi/bind/backends: fix EstimateGas * accounts, console, internal, rpc: minor error interface cleanups * Revert "accounts, console, internal, rpc: minor error interface cleanups" This reverts commit 2d3ef53c5304e429a04983210a417c1f4e0dafb7. * re-apply the good parts of 2d3ef53c53 * rpc: add test for returning server error data from client Co-authored-by: rjl493456442 <garyrong0905@gmail.com> Co-authored-by: Péter Szilágyi <peterke@gmail.com> Co-authored-by: Felix Lange <fjl@twurst.com>
2020-06-08 03:09:49 -05:00
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
var resp interface{}
err := client.Call(&resp, "test_returnError")
if err == nil {
t.Fatal("expected error")
}
// Check code.
// The method handler returns an error value which implements the rpc.Error
// interface, i.e. it has a custom error code. The server returns this error code.
expectedCode := testError{}.ErrorCode()
internal/ethapi: return revert reason for eth_call (#21083) * internal/ethapi: return revert reason for eth_call * internal/ethapi: moved revert reason logic to doCall * accounts/abi/bind/backends: added revert reason logic to simulated backend * internal/ethapi: fixed linting error * internal/ethapi: check if require reason can be unpacked * internal/ethapi: better error logic * internal/ethapi: simplify logic * internal/ethapi: return vmError() * internal/ethapi: move handling of revert out of docall * graphql: removed revert logic until spec change * rpc: internal/ethapi: added custom error types * graphql: use returndata instead of return Return() checks if there is an error. If an error is found, we return nil. For most use cases it can be beneficial to return the output even if there was an error. This code should be changed anyway once the spec supports error reasons in graphql responses * accounts/abi/bind/backends: added tests for revert reason * internal/ethapi: add errorCode to revert error * internal/ethapi: add errorCode of 3 to revertError * internal/ethapi: unified estimateGasErrors, simplified logic * internal/ethapi: unified handling of errors in DoEstimateGas * rpc: print error data field * accounts/abi/bind/backends: unify simulatedBackend and RPC * internal/ethapi: added binary data to revertError data * internal/ethapi: refactored unpacking logic into newRevertError * accounts/abi/bind/backends: fix EstimateGas * accounts, console, internal, rpc: minor error interface cleanups * Revert "accounts, console, internal, rpc: minor error interface cleanups" This reverts commit 2d3ef53c5304e429a04983210a417c1f4e0dafb7. * re-apply the good parts of 2d3ef53c53 * rpc: add test for returning server error data from client Co-authored-by: rjl493456442 <garyrong0905@gmail.com> Co-authored-by: Péter Szilágyi <peterke@gmail.com> Co-authored-by: Felix Lange <fjl@twurst.com>
2020-06-08 03:09:49 -05:00
if e, ok := err.(Error); !ok {
t.Fatalf("client did not return rpc.Error, got %#v", e)
} else if e.ErrorCode() != expectedCode {
t.Fatalf("wrong error code %d, want %d", e.ErrorCode(), expectedCode)
internal/ethapi: return revert reason for eth_call (#21083) * internal/ethapi: return revert reason for eth_call * internal/ethapi: moved revert reason logic to doCall * accounts/abi/bind/backends: added revert reason logic to simulated backend * internal/ethapi: fixed linting error * internal/ethapi: check if require reason can be unpacked * internal/ethapi: better error logic * internal/ethapi: simplify logic * internal/ethapi: return vmError() * internal/ethapi: move handling of revert out of docall * graphql: removed revert logic until spec change * rpc: internal/ethapi: added custom error types * graphql: use returndata instead of return Return() checks if there is an error. If an error is found, we return nil. For most use cases it can be beneficial to return the output even if there was an error. This code should be changed anyway once the spec supports error reasons in graphql responses * accounts/abi/bind/backends: added tests for revert reason * internal/ethapi: add errorCode to revert error * internal/ethapi: add errorCode of 3 to revertError * internal/ethapi: unified estimateGasErrors, simplified logic * internal/ethapi: unified handling of errors in DoEstimateGas * rpc: print error data field * accounts/abi/bind/backends: unify simulatedBackend and RPC * internal/ethapi: added binary data to revertError data * internal/ethapi: refactored unpacking logic into newRevertError * accounts/abi/bind/backends: fix EstimateGas * accounts, console, internal, rpc: minor error interface cleanups * Revert "accounts, console, internal, rpc: minor error interface cleanups" This reverts commit 2d3ef53c5304e429a04983210a417c1f4e0dafb7. * re-apply the good parts of 2d3ef53c53 * rpc: add test for returning server error data from client Co-authored-by: rjl493456442 <garyrong0905@gmail.com> Co-authored-by: Péter Szilágyi <peterke@gmail.com> Co-authored-by: Felix Lange <fjl@twurst.com>
2020-06-08 03:09:49 -05:00
}
internal/ethapi: return revert reason for eth_call (#21083) * internal/ethapi: return revert reason for eth_call * internal/ethapi: moved revert reason logic to doCall * accounts/abi/bind/backends: added revert reason logic to simulated backend * internal/ethapi: fixed linting error * internal/ethapi: check if require reason can be unpacked * internal/ethapi: better error logic * internal/ethapi: simplify logic * internal/ethapi: return vmError() * internal/ethapi: move handling of revert out of docall * graphql: removed revert logic until spec change * rpc: internal/ethapi: added custom error types * graphql: use returndata instead of return Return() checks if there is an error. If an error is found, we return nil. For most use cases it can be beneficial to return the output even if there was an error. This code should be changed anyway once the spec supports error reasons in graphql responses * accounts/abi/bind/backends: added tests for revert reason * internal/ethapi: add errorCode to revert error * internal/ethapi: add errorCode of 3 to revertError * internal/ethapi: unified estimateGasErrors, simplified logic * internal/ethapi: unified handling of errors in DoEstimateGas * rpc: print error data field * accounts/abi/bind/backends: unify simulatedBackend and RPC * internal/ethapi: added binary data to revertError data * internal/ethapi: refactored unpacking logic into newRevertError * accounts/abi/bind/backends: fix EstimateGas * accounts, console, internal, rpc: minor error interface cleanups * Revert "accounts, console, internal, rpc: minor error interface cleanups" This reverts commit 2d3ef53c5304e429a04983210a417c1f4e0dafb7. * re-apply the good parts of 2d3ef53c53 * rpc: add test for returning server error data from client Co-authored-by: rjl493456442 <garyrong0905@gmail.com> Co-authored-by: Péter Szilágyi <peterke@gmail.com> Co-authored-by: Felix Lange <fjl@twurst.com>
2020-06-08 03:09:49 -05:00
// Check data.
if e, ok := err.(DataError); !ok {
t.Fatalf("client did not return rpc.DataError, got %#v", e)
} else if e.ErrorData() != (testError{}.ErrorData()) {
t.Fatalf("wrong error data %#v, want %#v", e.ErrorData(), testError{}.ErrorData())
}
}
func TestClientBatchRequest(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
batch := []BatchElem{
{
Method: "test_echo",
Args: []interface{}{"hello", 10, &echoArgs{"world"}},
Result: new(echoResult),
},
{
Method: "test_echo",
Args: []interface{}{"hello2", 11, &echoArgs{"world"}},
Result: new(echoResult),
},
{
Method: "no_such_method",
Args: []interface{}{1, 2, 3},
Result: new(int),
},
}
if err := client.BatchCall(batch); err != nil {
t.Fatal(err)
}
wantResult := []BatchElem{
{
Method: "test_echo",
Args: []interface{}{"hello", 10, &echoArgs{"world"}},
Result: &echoResult{"hello", 10, &echoArgs{"world"}},
},
{
Method: "test_echo",
Args: []interface{}{"hello2", 11, &echoArgs{"world"}},
Result: &echoResult{"hello2", 11, &echoArgs{"world"}},
},
{
Method: "no_such_method",
Args: []interface{}{1, 2, 3},
Result: new(int),
Error: &jsonError{Code: -32601, Message: "the method no_such_method does not exist/is not available"},
},
}
if !reflect.DeepEqual(batch, wantResult) {
t.Errorf("batch results mismatch:\ngot %swant %s", spew.Sdump(batch), spew.Sdump(wantResult))
}
}
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
// This checks that, for HTTP connections, the length of batch responses is validated to
// match the request exactly.
func TestClientBatchRequest_len(t *testing.T) {
t.Parallel()
b, err := json.Marshal([]jsonrpcMessage{
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
{Version: "2.0", ID: json.RawMessage("1"), Result: json.RawMessage(`"0x1"`)},
{Version: "2.0", ID: json.RawMessage("2"), Result: json.RawMessage(`"0x2"`)},
})
if err != nil {
t.Fatal("failed to encode jsonrpc message:", err)
}
s := httptest.NewServer(http.HandlerFunc(func(rw http.ResponseWriter, r *http.Request) {
_, err := rw.Write(b)
if err != nil {
t.Error("failed to write response:", err)
}
}))
t.Cleanup(s.Close)
t.Run("too-few", func(t *testing.T) {
t.Parallel()
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
client, err := Dial(s.URL)
if err != nil {
t.Fatal("failed to dial test server:", err)
}
defer client.Close()
batch := []BatchElem{
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
{Method: "foo", Result: new(string)},
{Method: "bar", Result: new(string)},
{Method: "baz", Result: new(string)},
}
ctx, cancelFn := context.WithTimeout(context.Background(), time.Second)
defer cancelFn()
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
if err := client.BatchCallContext(ctx, batch); err != nil {
t.Fatal("error:", err)
}
for i, elem := range batch[:2] {
if elem.Error != nil {
t.Errorf("expected no error for batch element %d, got %q", i, elem.Error)
}
}
for i, elem := range batch[2:] {
if elem.Error != ErrMissingBatchResponse {
t.Errorf("wrong error %q for batch element %d", elem.Error, i+2)
}
}
})
t.Run("too-many", func(t *testing.T) {
t.Parallel()
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
client, err := Dial(s.URL)
if err != nil {
t.Fatal("failed to dial test server:", err)
}
defer client.Close()
batch := []BatchElem{
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
{Method: "foo", Result: new(string)},
}
ctx, cancelFn := context.WithTimeout(context.Background(), time.Second)
defer cancelFn()
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
if err := client.BatchCallContext(ctx, batch); err != nil {
t.Fatal("error:", err)
}
for i, elem := range batch[:1] {
if elem.Error != nil {
t.Errorf("expected no error for batch element %d, got %q", i, elem.Error)
}
}
for i, elem := range batch[1:] {
if elem.Error != ErrMissingBatchResponse {
t.Errorf("wrong error %q for batch element %d", elem.Error, i+2)
}
}
})
}
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
// This checks that the client can handle the case where the server doesn't
// respond to all requests in a batch.
func TestClientBatchRequestLimit(t *testing.T) {
t.Parallel()
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
server := newTestServer()
defer server.Stop()
server.SetBatchLimits(2, 100000)
client := DialInProc(server)
2024-04-16 03:38:25 -05:00
defer client.Close()
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
batch := []BatchElem{
{Method: "foo"},
{Method: "bar"},
{Method: "baz"},
}
err := client.BatchCall(batch)
if err != nil {
t.Fatal("unexpected error:", err)
}
// Check that the first response indicates an error with batch size.
var err0 Error
if !errors.As(batch[0].Error, &err0) {
t.Log("error zero:", batch[0].Error)
t.Fatalf("batch elem 0 has wrong error type: %T", batch[0].Error)
} else {
if err0.ErrorCode() != -32600 || err0.Error() != errMsgBatchTooLarge {
t.Fatalf("wrong error on batch elem zero: %v", err0)
}
}
// Check that remaining response batch elements are reported as absent.
for i, elem := range batch[1:] {
if elem.Error != ErrMissingBatchResponse {
t.Fatalf("batch elem %d has unexpected error: %v", i+1, elem.Error)
}
}
}
func TestClientNotify(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
if err := client.Notify(context.Background(), "test_echo", "hello", 10, &echoArgs{"world"}); err != nil {
t.Fatal(err)
}
}
// func TestClientCancelInproc(t *testing.T) { testClientCancel("inproc", t) }
func TestClientCancelWebsocket(t *testing.T) { testClientCancel("ws", t) }
func TestClientCancelHTTP(t *testing.T) { testClientCancel("http", t) }
func TestClientCancelIPC(t *testing.T) { testClientCancel("ipc", t) }
// This test checks that requests made through CallContext can be canceled by canceling
// the context.
func testClientCancel(transport string, t *testing.T) {
// These tests take a lot of time, run them all at once.
// You probably want to run with -parallel 1 or comment out
// the call to t.Parallel if you enable the logging.
t.Parallel()
server := newTestServer()
defer server.Stop()
// What we want to achieve is that the context gets canceled
// at various stages of request processing. The interesting cases
// are:
// - cancel during dial
// - cancel while performing a HTTP request
// - cancel while waiting for a response
//
// To trigger those, the times are chosen such that connections
// are killed within the deadline for every other call (maxKillTimeout
// is 2x maxCancelTimeout).
//
// Once a connection is dead, there is a fair chance it won't connect
// successfully because the accept is delayed by 1s.
maxContextCancelTimeout := 300 * time.Millisecond
fl := &flakeyListener{
maxAcceptDelay: 1 * time.Second,
maxKillTimeout: 600 * time.Millisecond,
}
var client *Client
switch transport {
case "ws", "http":
c, hs := httpTestClient(server, transport, fl)
defer hs.Close()
client = c
case "ipc":
c, l := ipcTestClient(server, fl)
defer l.Close()
client = c
default:
panic("unknown transport: " + transport)
}
2024-04-16 03:38:25 -05:00
defer client.Close()
// The actual test starts here.
var (
wg sync.WaitGroup
nreqs = 10
ncallers = 10
)
caller := func(index int) {
defer wg.Done()
for i := 0; i < nreqs; i++ {
var (
ctx context.Context
cancel func()
timeout = time.Duration(rand.Int63n(int64(maxContextCancelTimeout)))
)
if index < ncallers/2 {
// For half of the callers, create a context without deadline
// and cancel it later.
ctx, cancel = context.WithCancel(context.Background())
time.AfterFunc(timeout, cancel)
} else {
// For the other half, create a context with a deadline instead. This is
// different because the context deadline is used to set the socket write
// deadline.
ctx, cancel = context.WithTimeout(context.Background(), timeout)
}
// Now perform a call with the context.
// The key thing here is that no call will ever complete successfully.
err := client.CallContext(ctx, nil, "test_block")
switch {
case err == nil:
_, hasDeadline := ctx.Deadline()
t.Errorf("no error for call with %v wait time (deadline: %v)", timeout, hasDeadline)
// default:
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
// t.Logf("got expected error with %v wait time: %v", timeout, err)
}
cancel()
}
}
wg.Add(ncallers)
for i := 0; i < ncallers; i++ {
go caller(i)
}
wg.Wait()
}
func TestClientSubscribeInvalidArg(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
check := func(shouldPanic bool, arg interface{}) {
defer func() {
err := recover()
if shouldPanic && err == nil {
t.Errorf("EthSubscribe should've panicked for %#v", arg)
}
if !shouldPanic && err != nil {
t.Errorf("EthSubscribe shouldn't have panicked for %#v", arg)
buf := make([]byte, 1024*1024)
buf = buf[:runtime.Stack(buf, false)]
t.Error(err)
t.Error(string(buf))
}
}()
client.EthSubscribe(context.Background(), arg, "foo_bar")
}
check(true, nil)
check(true, 1)
check(true, (chan int)(nil))
check(true, make(<-chan int))
check(false, make(chan int))
check(false, make(chan<- int))
}
func TestClientSubscribe(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client := DialInProc(server)
defer client.Close()
nc := make(chan int)
count := 10
sub, err := client.Subscribe(context.Background(), "nftest", nc, "someSubscription", count, 0)
if err != nil {
t.Fatal("can't subscribe:", err)
}
for i := 0; i < count; i++ {
if val := <-nc; val != i {
t.Fatalf("value mismatch: got %d, want %d", val, i)
}
}
sub.Unsubscribe()
select {
case v := <-nc:
t.Fatal("received value after unsubscribe:", v)
case err := <-sub.Err():
if err != nil {
t.Fatalf("Err returned a non-nil error after explicit unsubscribe: %q", err)
}
case <-time.After(1 * time.Second):
t.Fatalf("subscription not closed within 1s after unsubscribe")
}
}
// In this test, the connection drops while Subscribe is waiting for a response.
func TestClientSubscribeClose(t *testing.T) {
t.Parallel()
server := newTestServer()
service := &notificationTestService{
gotHangSubscriptionReq: make(chan struct{}),
unblockHangSubscription: make(chan struct{}),
}
if err := server.RegisterName("nftest2", service); err != nil {
t.Fatal(err)
}
defer server.Stop()
client := DialInProc(server)
defer client.Close()
var (
nc = make(chan int)
errc = make(chan error, 1)
sub *ClientSubscription
err error
)
go func() {
sub, err = client.Subscribe(context.Background(), "nftest2", nc, "hangSubscription", 999)
errc <- err
}()
<-service.gotHangSubscriptionReq
client.Close()
service.unblockHangSubscription <- struct{}{}
select {
case err := <-errc:
if err == nil {
t.Errorf("Subscribe returned nil error after Close")
}
if sub != nil {
t.Error("Subscribe returned non-nil subscription after Close")
}
case <-time.After(1 * time.Second):
t.Fatalf("Subscribe did not return within 1s after Close")
}
}
// This test reproduces https://github.com/ethereum/go-ethereum/issues/17837 where the
// client hangs during shutdown when Unsubscribe races with Client.Close.
func TestClientCloseUnsubscribeRace(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
for i := 0; i < 20; i++ {
client := DialInProc(server)
nc := make(chan int)
sub, err := client.Subscribe(context.Background(), "nftest", nc, "someSubscription", 3, 1)
if err != nil {
t.Fatal(err)
}
go client.Close()
go sub.Unsubscribe()
select {
case <-sub.Err():
case <-time.After(5 * time.Second):
t.Fatal("subscription not closed within timeout")
}
}
}
// unsubscribeBlocker will wait for the quit channel to process an unsubscribe
// request.
type unsubscribeBlocker struct {
ServerCodec
quit chan struct{}
}
func (b *unsubscribeBlocker) readBatch() ([]*jsonrpcMessage, bool, error) {
msgs, batch, err := b.ServerCodec.readBatch()
for _, msg := range msgs {
if msg.isUnsubscribe() {
<-b.quit
}
}
return msgs, batch, err
}
// TestUnsubscribeTimeout verifies that calling the client's Unsubscribe
// function will eventually timeout and not block forever in case the serve does
// not respond.
// It reproducers the issue https://github.com/ethereum/go-ethereum/issues/30156
func TestUnsubscribeTimeout(t *testing.T) {
t.Parallel()
srv := NewServer()
srv.RegisterName("nftest", new(notificationTestService))
// Setup middleware to block on unsubscribe.
p1, p2 := net.Pipe()
blocker := &unsubscribeBlocker{ServerCodec: NewCodec(p1), quit: make(chan struct{})}
defer close(blocker.quit)
// Serve the middleware.
go srv.ServeCodec(blocker, OptionMethodInvocation|OptionSubscriptions)
defer srv.Stop()
// Create the client on the other end of the pipe.
cfg := new(clientConfig)
client, _ := newClient(context.Background(), cfg, func(context.Context) (ServerCodec, error) {
return NewCodec(p2), nil
})
defer client.Close()
// Start subscription.
sub, err := client.Subscribe(context.Background(), "nftest", make(chan int), "someSubscription", 1, 1)
if err != nil {
t.Fatalf("failed to subscribe: %v", err)
}
// Now on a separate thread, attempt to unsubscribe. Since the middleware
// won't return, the function will only return if it times out on the request.
done := make(chan struct{})
go func() {
sub.Unsubscribe()
done <- struct{}{}
}()
// Wait for the timeout. If the expected time for the timeout elapses, the
// test is considered failed.
select {
case <-done:
case <-time.After(unsubscribeTimeout + 3*time.Second):
t.Fatalf("Unsubscribe did not return within %s", unsubscribeTimeout)
}
}
// unsubscribeRecorder collects the subscription IDs of *_unsubscribe calls.
type unsubscribeRecorder struct {
ServerCodec
unsubscribes map[string]bool
}
func (r *unsubscribeRecorder) readBatch() ([]*jsonrpcMessage, bool, error) {
if r.unsubscribes == nil {
r.unsubscribes = make(map[string]bool)
}
msgs, batch, err := r.ServerCodec.readBatch()
for _, msg := range msgs {
if msg.isUnsubscribe() {
var params []string
if err := json.Unmarshal(msg.Params, &params); err != nil {
panic("unsubscribe decode error: " + err.Error())
}
r.unsubscribes[params[0]] = true
}
}
return msgs, batch, err
}
// This checks that Client calls the _unsubscribe method on the server when Unsubscribe is
// called on a subscription.
func TestClientSubscriptionUnsubscribeServer(t *testing.T) {
t.Parallel()
// Create the server.
srv := NewServer()
srv.RegisterName("nftest", new(notificationTestService))
p1, p2 := net.Pipe()
recorder := &unsubscribeRecorder{ServerCodec: NewCodec(p1)}
go srv.ServeCodec(recorder, OptionMethodInvocation|OptionSubscriptions)
defer srv.Stop()
// Create the client on the other end of the pipe.
rpc: add limit for batch request items and response size (#26681) This PR adds server-side limits for JSON-RPC batch requests. Before this change, batches were limited only by processing time. The server would pick calls from the batch and answer them until the response timeout occurred, then stop processing the remaining batch items. Here, we are adding two additional limits which can be configured: - the 'item limit': batches can have at most N items - the 'response size limit': batches can contain at most X response bytes These limits are optional in package rpc. In Geth, we set a default limit of 1000 items and 25MB response size. When a batch goes over the limit, an error response is returned to the client. However, doing this correctly isn't always possible. In JSON-RPC, only method calls with a valid `id` can be responded to. Since batches may also contain non-call messages or notifications, the best effort thing we can do to report an error with the batch itself is reporting the limit violation as an error for the first method call in the batch. If a batch is too large, but contains only notifications and responses, the error will be reported with a null `id`. The RPC client was also changed so it can deal with errors resulting from too large batches. An older client connected to the server code in this PR could get stuck until the request timeout occurred when the batch is too large. **Upgrading to a version of the RPC client containing this change is strongly recommended to avoid timeout issues.** For some weird reason, when writing the original client implementation, @fjl worked off of the assumption that responses could be distributed across batches arbitrarily. So for a batch request containing requests `[A B C]`, the server could respond with `[A B C]` but also with `[A B] [C]` or even `[A] [B] [C]` and it wouldn't make a difference to the client. So in the implementation of BatchCallContext, the client waited for all requests in the batch individually. If the server didn't respond to some of the requests in the batch, the client would eventually just time out (if a context was used). With the addition of batch limits into the server, we anticipate that people will hit this kind of error way more often. To handle this properly, the client now waits for a single response batch and expects it to contain all responses to the requests. --------- Co-authored-by: Felix Lange <fjl@twurst.com> Co-authored-by: Martin Holst Swende <martin@swende.se>
2023-06-13 06:38:58 -05:00
cfg := new(clientConfig)
client, _ := newClient(context.Background(), cfg, func(context.Context) (ServerCodec, error) {
return NewCodec(p2), nil
})
defer client.Close()
// Create the subscription.
ch := make(chan int)
sub, err := client.Subscribe(context.Background(), "nftest", ch, "someSubscription", 1, 1)
if err != nil {
t.Fatal(err)
}
// Unsubscribe and check that unsubscribe was called.
sub.Unsubscribe()
if !recorder.unsubscribes[sub.subid] {
t.Fatal("client did not call unsubscribe method")
}
if _, open := <-sub.Err(); open {
t.Fatal("subscription error channel not closed after unsubscribe")
}
}
// This checks that the subscribed channel can be closed after Unsubscribe.
// It is the reproducer for https://github.com/ethereum/go-ethereum/issues/22322
func TestClientSubscriptionChannelClose(t *testing.T) {
t.Parallel()
var (
srv = NewServer()
httpsrv = httptest.NewServer(srv.WebsocketHandler(nil))
wsURL = "ws:" + strings.TrimPrefix(httpsrv.URL, "http:")
)
defer srv.Stop()
defer httpsrv.Close()
srv.RegisterName("nftest", new(notificationTestService))
client, _ := Dial(wsURL)
2024-04-16 03:38:25 -05:00
defer client.Close()
for i := 0; i < 100; i++ {
ch := make(chan int, 100)
sub, err := client.Subscribe(context.Background(), "nftest", ch, "someSubscription", 100, 1)
if err != nil {
t.Fatal(err)
}
sub.Unsubscribe()
close(ch)
}
}
// This test checks that Client doesn't lock up when a single subscriber
// doesn't read subscription events.
func TestClientNotificationStorm(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
doTest := func(count int, wantError bool) {
client := DialInProc(server)
defer client.Close()
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
defer cancel()
// Subscribe on the server. It will start sending many notifications
// very quickly.
nc := make(chan int)
sub, err := client.Subscribe(ctx, "nftest", nc, "someSubscription", count, 0)
if err != nil {
t.Fatal("can't subscribe:", err)
}
defer sub.Unsubscribe()
// Process each notification, try to run a call in between each of them.
for i := 0; i < count; i++ {
select {
case val := <-nc:
if val != i {
t.Fatalf("(%d/%d) unexpected value %d", i, count, val)
}
case err := <-sub.Err():
if wantError && err != ErrSubscriptionQueueOverflow {
t.Fatalf("(%d/%d) got error %q, want %q", i, count, err, ErrSubscriptionQueueOverflow)
} else if !wantError {
t.Fatalf("(%d/%d) got unexpected error %q", i, count, err)
}
return
}
var r int
err := client.CallContext(ctx, &r, "nftest_echo", i)
if err != nil {
if !wantError {
t.Fatalf("(%d/%d) call error: %v", i, count, err)
}
return
}
}
if wantError {
t.Fatalf("didn't get expected error")
}
}
doTest(8000, false)
doTest(24000, true)
}
func TestClientSetHeader(t *testing.T) {
t.Parallel()
var gotHeader bool
srv := newTestServer()
httpsrv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
if r.Header.Get("test") == "ok" {
gotHeader = true
}
srv.ServeHTTP(w, r)
}))
defer httpsrv.Close()
defer srv.Stop()
client, err := Dial(httpsrv.URL)
if err != nil {
t.Fatal(err)
}
defer client.Close()
client.SetHeader("test", "ok")
if _, err := client.SupportedModules(); err != nil {
t.Fatal(err)
}
if !gotHeader {
t.Fatal("client did not set custom header")
}
// Check that Content-Type can be replaced.
client.SetHeader("content-type", "application/x-garbage")
_, err = client.SupportedModules()
if err == nil {
t.Fatal("no error for invalid content-type header")
} else if !strings.Contains(err.Error(), "Unsupported Media Type") {
t.Fatalf("error is not related to content-type: %q", err)
}
}
func TestClientHTTP(t *testing.T) {
t.Parallel()
server := newTestServer()
defer server.Stop()
client, hs := httpTestClient(server, "http", nil)
defer hs.Close()
defer client.Close()
// Launch concurrent requests.
var (
results = make([]echoResult, 100)
errc = make(chan error, len(results))
wantResult = echoResult{"a", 1, new(echoArgs)}
)
for i := range results {
go func() {
errc <- client.Call(&results[i], "test_echo", wantResult.String, wantResult.Int, wantResult.Args)
}()
}
// Wait for all of them to complete.
timeout := time.NewTimer(5 * time.Second)
defer timeout.Stop()
for i := range results {
select {
case err := <-errc:
if err != nil {
t.Fatal(err)
}
case <-timeout.C:
t.Fatalf("timeout (got %d/%d) results)", i+1, len(results))
}
}
// Check results.
for i := range results {
if !reflect.DeepEqual(results[i], wantResult) {
t.Errorf("result %d mismatch: got %#v, want %#v", i, results[i], wantResult)
}
}
}
func TestClientReconnect(t *testing.T) {
t.Parallel()
startServer := func(addr string) (*Server, net.Listener) {
srv := newTestServer()
l, err := net.Listen("tcp", addr)
if err != nil {
t.Fatal("can't listen:", err)
}
go http.Serve(l, srv.WebsocketHandler([]string{"*"}))
return srv, l
}
ctx, cancel := context.WithTimeout(context.Background(), 12*time.Second)
defer cancel()
// Start a server and corresponding client.
s1, l1 := startServer("127.0.0.1:0")
client, err := DialContext(ctx, "ws://"+l1.Addr().String())
if err != nil {
t.Fatal("can't dial", err)
}
defer client.Close()
// Perform a call. This should work because the server is up.
var resp echoResult
if err := client.CallContext(ctx, &resp, "test_echo", "", 1, nil); err != nil {
t.Fatal(err)
}
// Shut down the server and allow for some cool down time so we can listen on the same
// address again.
l1.Close()
s1.Stop()
time.Sleep(2 * time.Second)
// Try calling again. It shouldn't work.
if err := client.CallContext(ctx, &resp, "test_echo", "", 2, nil); err == nil {
t.Error("successful call while the server is down")
t.Logf("resp: %#v", resp)
}
// Start it up again and call again. The connection should be reestablished.
// We spawn multiple calls here to check whether this hangs somehow.
s2, l2 := startServer(l1.Addr().String())
defer l2.Close()
defer s2.Stop()
start := make(chan struct{})
errors := make(chan error, 20)
for i := 0; i < cap(errors); i++ {
go func() {
<-start
var resp echoResult
errors <- client.CallContext(ctx, &resp, "test_echo", "", 3, nil)
}()
}
close(start)
errcount := 0
for i := 0; i < cap(errors); i++ {
if err = <-errors; err != nil {
errcount++
}
}
t.Logf("%d errors, last error: %v", errcount, err)
if errcount > 1 {
t.Errorf("expected one error after disconnect, got %d", errcount)
}
}
func httpTestClient(srv *Server, transport string, fl *flakeyListener) (*Client, *httptest.Server) {
// Create the HTTP server.
var hs *httptest.Server
switch transport {
case "ws":
hs = httptest.NewUnstartedServer(srv.WebsocketHandler([]string{"*"}))
case "http":
hs = httptest.NewUnstartedServer(srv)
default:
panic("unknown HTTP transport: " + transport)
}
// Wrap the listener if required.
if fl != nil {
fl.Listener = hs.Listener
hs.Listener = fl
}
// Connect the client.
hs.Start()
client, err := Dial(transport + "://" + hs.Listener.Addr().String())
if err != nil {
panic(err)
}
return client, hs
}
func ipcTestClient(srv *Server, fl *flakeyListener) (*Client, net.Listener) {
// Listen on a random endpoint.
endpoint := fmt.Sprintf("go-ethereum-test-ipc-%d-%d", os.Getpid(), rand.Int63())
if runtime.GOOS == "windows" {
endpoint = `\\.\pipe\` + endpoint
} else {
endpoint = os.TempDir() + "/" + endpoint
}
l, err := ipcListen(endpoint)
if err != nil {
panic(err)
}
// Connect the listener to the server.
if fl != nil {
fl.Listener = l
l = fl
}
go srv.ServeListener(l)
// Connect the client.
client, err := Dial(endpoint)
if err != nil {
panic(err)
}
return client, l
}
// flakeyListener kills accepted connections after a random timeout.
type flakeyListener struct {
net.Listener
maxKillTimeout time.Duration
maxAcceptDelay time.Duration
}
func (l *flakeyListener) Accept() (net.Conn, error) {
delay := time.Duration(rand.Int63n(int64(l.maxAcceptDelay)))
time.Sleep(delay)
c, err := l.Listener.Accept()
if err == nil {
timeout := time.Duration(rand.Int63n(int64(l.maxKillTimeout)))
time.AfterFunc(timeout, func() {
log.Debug(fmt.Sprintf("killing conn %v after %v", c.LocalAddr(), timeout))
c.Close()
})
}
return c, err
}