Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[#1651] improvement(netty): Set Netty as the default server type #1919

Draft
wants to merge 3 commits into
base: master
Choose a base branch
from
Draft

[#1651] improvement(netty): Set Netty as the default server type #1919

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
f618cb7
[#1651] improvement(netty): Set Netty as the default server type
rickyma Jul 17, 2024
d8512aa
[#1651] improvement(netty): Set Netty as the default server type
rickyma Jul 17, 2024
7045e72
[#1651] improvement(netty): Set Netty as the default server type
rickyma Jul 17, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions common/src/main/java/org/apache/uniffle/common/config/RssBaseConf.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,9 +36,9 @@ public class RssBaseConf extends RssConf {
public static final ConfigOption<ServerType> RPC_SERVER_TYPE =
ConfigOptions.key("rss.rpc.server.type")
.enumType(ServerType.class)
.defaultValue(ServerType.GRPC)
.defaultValue(ServerType.GRPC_NETTY)
.withDescription(
"Shuffle server type, supports GRPC_NETTY, GRPC. The default value is GRPC for now. But we recommend using GRPC_NETTY to enable Netty on the server side for better stability and performance.");
"Shuffle server type, supports GRPC_NETTY, GRPC. The default value is GRPC_NETTY. We recommend using GRPC_NETTY to enable Netty on the server side for better stability and performance.");

public static final ConfigOption<Integer> RPC_SERVER_PORT =
ConfigOptions.key("rss.rpc.server.port")
Expand Down
4 changes: 2 additions & 2 deletions common/src/main/java/org/apache/uniffle/common/config/RssClientConf.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -233,9 +233,9 @@ public class RssClientConf {
public static final ConfigOption<ClientType> RSS_CLIENT_TYPE =
ConfigOptions.key("rss.client.type")
.enumType(ClientType.class)
.defaultValue(ClientType.GRPC)
.defaultValue(ClientType.GRPC_NETTY)
.withDescription(
"Supports GRPC_NETTY, GRPC. The default value is GRPC. But we recommend using GRPC_NETTY to enable Netty on the client side for better stability and performance.");
"Supports GRPC_NETTY, GRPC. The default value is GRPC_NETTY. We recommend using GRPC_NETTY to enable Netty on the client side for better stability and performance.");

public static final ConfigOption<Boolean> RSS_CLIENT_REMOTE_STORAGE_USE_LOCAL_CONF_ENABLED =
ConfigOptions.key("rss.client.remote.storage.useLocalConfAsDefault")
Expand Down
2 changes: 1 addition & 1 deletion coordinator/src/test/java/org/apache/uniffle/coordinator/CoordinatorConfTest.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -34,7 +34,7 @@ public void test() {

// test base conf
assertEquals(9527, conf.getInteger(CoordinatorConf.RPC_SERVER_PORT));
assertEquals("GRPC", conf.get(CoordinatorConf.RPC_SERVER_TYPE).name());
assertEquals("GRPC_NETTY", conf.get(CoordinatorConf.RPC_SERVER_TYPE).name());
assertEquals(9526, conf.getInteger(CoordinatorConf.JETTY_HTTP_PORT));

// test coordinator specific conf
Expand Down
2 changes: 1 addition & 1 deletion coordinator/src/test/resources/coordinator.conf
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@

# base conf
rss.rpc.server.port 9527
rss.rpc.server.type GRPC
rss.rpc.server.type GRPC_NETTY
rss.jetty.http.port 9526

# coordinator specific conf
Expand Down
24 changes: 12 additions & 12 deletions docs/client_guide/client_guide.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -100,15 +100,15 @@ spark.rss.data.replica.read 2
```

### Netty Setting
| Property Name | Default | Description |
|----------------------------------------------------------------|---------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <client_type>.rss.client.type | GRPC | Supports GRPC_NETTY, GRPC. The default value is GRPC. But we recommend using GRPC_NETTY to enable Netty on the client side for better stability and performance. |
| <client_type>.rss.client.netty.io.mode | NIO | Netty EventLoopGroup backend, available options: NIO, EPOLL. |
| <client_type>.rss.client.netty.client.connection.timeout.ms | 600000 | Connection active timeout. |
| <client_type>.rss.client.netty.client.threads | 0 | Number of threads used in the client thread pool. Default is 0, Netty will use the number of (available logical cores * 2) as the number of threads. |
| <client_type>.rss.client.netty.client.prefer.direct.bufs | true | If true, we will prefer allocating off-heap byte buffers within Netty. |
| <client_type>.rss.client.netty.client.pooled.allocator.enabled | true | If true, we will use PooledByteBufAllocator to allocate byte buffers within Netty, otherwise we'll use UnpooledByteBufAllocator. |
| <client_type>.rss.client.netty.client.shared.allocator.enabled | true | A flag indicating whether to share the ByteBuf allocators between the different Netty channels when enabling Netty. If enabled then only three ByteBuf allocators are created: one PooledByteBufAllocator where caching is allowed, one PooledByteBufAllocator where not and one UnpooledByteBufAllocator. When disabled, a new allocator is created for each transport client. |
| <client_type>.rss.client.netty.client.connections.per.peer | 2 | Suppose there are 100 executors, spark.rss.client.netty.client.connections.per.peer = 2, then each ShuffleServer will establish a total of (100 * 2) connections with multiple clients. |
| <client_type>.rss.client.netty.client.receive.buffer | 0 | Receive buffer size (SO_RCVBUF). Note: the optimal size for receive buffer and send buffer should be latency * network_bandwidth. Assuming latency = 1ms, network_bandwidth = 10Gbps, buffer size should be ~ 1.25MB. Default is 0, the operating system automatically estimates the receive buffer size based on default settings. |
| <client_type>.rss.client.netty.client.send.buffer | 0 | Send buffer size (SO_SNDBUF). |
| Property Name | Default | Description |
|----------------------------------------------------------------|------------|---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
| <client_type>.rss.client.type | GRPC_NETTY | Supports GRPC_NETTY, GRPC. The default value is GRPC_NETTY. We recommend using GRPC_NETTY to enable Netty on the client side for better stability and performance. |
| <client_type>.rss.client.netty.io.mode | NIO | Netty EventLoopGroup backend, available options: NIO, EPOLL. |
| <client_type>.rss.client.netty.client.connection.timeout.ms | 600000 | Connection active timeout. |
| <client_type>.rss.client.netty.client.threads | 0 | Number of threads used in the client thread pool. Default is 0, Netty will use the number of (available logical cores * 2) as the number of threads. |
| <client_type>.rss.client.netty.client.prefer.direct.bufs | true | If true, we will prefer allocating off-heap byte buffers within Netty. |
| <client_type>.rss.client.netty.client.pooled.allocator.enabled | true | If true, we will use PooledByteBufAllocator to allocate byte buffers within Netty, otherwise we'll use UnpooledByteBufAllocator. |
| <client_type>.rss.client.netty.client.shared.allocator.enabled | true | A flag indicating whether to share the ByteBuf allocators between the different Netty channels when enabling Netty. If enabled then only three ByteBuf allocators are created: one PooledByteBufAllocator where caching is allowed, one PooledByteBufAllocator where not and one UnpooledByteBufAllocator. When disabled, a new allocator is created for each transport client. |
| <client_type>.rss.client.netty.client.connections.per.peer | 2 | Suppose there are 100 executors, spark.rss.client.netty.client.connections.per.peer = 2, then each ShuffleServer will establish a total of (100 * 2) connections with multiple clients. |
| <client_type>.rss.client.netty.client.receive.buffer | 0 | Receive buffer size (SO_RCVBUF). Note: the optimal size for receive buffer and send buffer should be latency * network_bandwidth. Assuming latency = 1ms, network_bandwidth = 10Gbps, buffer size should be ~ 1.25MB. Default is 0, the operating system automatically estimates the receive buffer size based on default settings. |
| <client_type>.rss.client.netty.client.send.buffer | 0 | Send buffer size (SO_SNDBUF). |
Loading
Loading