kazusa
/
skywalking-java
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Improve agent grpc auto reconnect (#3240) 

* improve auto connect

* add config

* use int instread of atomic

* update format

* update doc

* update var name and doc

* fix
This commit is contained in:
Alvin 2019-08-11 12:30:17 +08:00 committed by 吴晟 Wu Sheng
parent a328a3bafc
commit 90d525b164
4 changed files with 18 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
apm-sniffer/apm-agent-core/src/main/java/org/apache/skywalking/apm/agent/core/conf/Config.java
View File

@ -100,6 +100,13 @@ public class Config {
* after receiving reset command
*/
public static int COOL_DOWN_THRESHOLD = 10;
/**
* Force reconnection period of grpc, based on grpc_channel_check_interval.
* If count of check grpc channel status more than this number.
* The channel check will call channel.getState(true) to requestConnection.
*/
public static long FORCE_RECONNECTION_PERIOD = 1;
}
public static class Collector {

6
apm-sniffer/apm-agent-core/src/main/java/org/apache/skywalking/apm/agent/core/remote/GRPCChannel.java
View File

@ -75,7 +75,11 @@ public class GRPCChannel {
}
public boolean isConnected() {
return originChannel.getState(false) == ConnectivityState.READY;
return isConnected(false);
}
public boolean isConnected(boolean requestConnection) {
return originChannel.getState(requestConnection) == ConnectivityState.READY;
}
public static class Builder {

8
apm-sniffer/apm-agent-core/src/main/java/org/apache/skywalking/apm/agent/core/remote/GRPCChannelManager.java
View File

@ -51,6 +51,7 @@ public class GRPCChannelManager implements BootService, Runnable {
private List<GRPCChannelListener> listeners = Collections.synchronizedList(new LinkedList<GRPCChannelListener>());
private volatile List<String> grpcServers;
private volatile int selectedIdx = -1;
private volatile int reconnectCount = 0;
@Override
public void prepare() throws Throwable {
@ -115,13 +116,14 @@ public class GRPCChannelManager implements BootService, Runnable {
.addChannelDecorator(new AgentIDDecorator())
.addChannelDecorator(new AuthenticationDecorator())
.build();
notify(GRPCChannelStatus.CONNECTED);
reconnectCount = 0;
reconnect = false;
} else if (managedChannel.isConnected()) {
} else if (managedChannel.isConnected(++reconnectCount > Config.Agent.FORCE_RECONNECTION_PERIOD)) {
// Reconnect to the same server is automatically done by GRPC,
// therefore we are responsible to check the connectivity and
// set the state and notify listeners
reconnectCount = 0;
notify(GRPCChannelStatus.CONNECTED);
reconnect = false;
}
@ -168,7 +170,7 @@ public class GRPCChannelManager implements BootService, Runnable {
private boolean isNetworkError(Throwable throwable) {
if (throwable instanceof StatusRuntimeException) {
StatusRuntimeException statusRuntimeException = (StatusRuntimeException)throwable;
StatusRuntimeException statusRuntimeException = (StatusRuntimeException) throwable;
return statusEquals(statusRuntimeException.getStatus(),
Status.UNAVAILABLE,
Status.PERMISSION_DENIED,

1
docs/en/setup/service-agent/java-agent/README.md
View File

@ -76,6 +76,7 @@ property key | Description | Default |
`agent.cause_exception_depth`|How depth the agent goes, when log all cause exceptions.|`5`|
`agent.active_v1_header `|Deactivate V1 header in default.|`false`|
`agent.cool_down_threshold `|How long should the agent wait (in minute) before re-registering to the OAP server after receiving reset command.|`10`|
`agent.force_reconnection_period `|Force reconnection period of grpc, based on grpc_channel_check_interval.|`1`|
`collector.grpc_channel_check_interval`|grpc channel status check interval.|`30`|
`collector.app_and_service_register_check_interval`|application and service registry check interval.|`3`|
`collector.backend_service`|Collector SkyWalking trace receiver service addresses.|`127.0.0.1:11800`|