GH-3199: TCP FailoverClientCF fail back default

Resolves https://github.com/spring-projects/spring-integration/issues/3199

- change the default behavior to not fail back until the current connection fails
- reduce method complexity (Sonar)
- 5.3 only
This commit is contained in:
Gary Russell
2020-03-02 14:22:42 -05:00
committed by Artem Bilan
parent c6cf261e25
commit cdbaf18664
2 changed files with 31 additions and 21 deletions

View File

@@ -42,7 +42,7 @@ import org.springframework.util.Assert;
*/
public class FailoverClientConnectionFactory extends AbstractClientConnectionFactory {
private static final long DEFAULT_REFRESH_SHARED_INTERVAL = 0L;
private static final long DEFAULT_REFRESH_SHARED_INTERVAL = Long.MAX_VALUE;
private final List<AbstractClientConnectionFactory> factories;
@@ -50,7 +50,9 @@ public class FailoverClientConnectionFactory extends AbstractClientConnectionFac
private long refreshSharedInterval = DEFAULT_REFRESH_SHARED_INTERVAL;
private boolean closeOnRefresh;
private boolean closeOnRefresh = true;
private boolean failBack;
private volatile long creationTime;
@@ -70,10 +72,9 @@ public class FailoverClientConnectionFactory extends AbstractClientConnectionFac
/**
* When using a shared connection {@link #setSingleUse(boolean) singleUse} is false,
* specify how long to wait before trying to fail back to start from the beginning of
* the factory list. Default is 0 for backwards compatibility to always try to get a
* connection to the primary server. If you don't want to fail back until the current
* connection is closed, set this to {@link Long#MAX_VALUE}.
* Cannot be changed when using {@link CachingClientConnectionFactory} delegates.
* the factory list. Default is {@link Long#MAX_VALUE} - meaning only fail back when
* the current connection fails. Cannot be changed when using
* {@link CachingClientConnectionFactory} delegates.
* @param refreshSharedInterval the interval in milliseconds.
* @since 4.3.22
* @see #setSingleUse(boolean)
@@ -83,6 +84,7 @@ public class FailoverClientConnectionFactory extends AbstractClientConnectionFac
Assert.isTrue(!this.cachingDelegates,
"'refreshSharedInterval' cannot be changed when using 'CachingClientConnectionFactory` delegates");
this.refreshSharedInterval = refreshSharedInterval;
this.failBack = refreshSharedInterval != Long.MAX_VALUE;
}
/**
@@ -149,7 +151,7 @@ public class FailoverClientConnectionFactory extends AbstractClientConnectionFac
protected TcpConnectionSupport obtainConnection() throws InterruptedException {
FailoverTcpConnection sharedConnection = (FailoverTcpConnection) getTheConnection();
boolean shared = !isSingleUse() && !this.cachingDelegates;
boolean refreshShared = shared
boolean refreshShared = this.failBack && shared
&& sharedConnection != null
&& System.currentTimeMillis() > this.creationTime + this.refreshSharedInterval;
if (sharedConnection != null && sharedConnection.isOpen() && !refreshShared) {
@@ -162,20 +164,26 @@ public class FailoverClientConnectionFactory extends AbstractClientConnectionFac
}
failoverTcpConnection.incrementEpoch();
if (shared) {
this.creationTime = System.currentTimeMillis();
/*
* We may have simply wrapped the same connection in a new wrapper; don't close.
*/
if (refreshShared && this.closeOnRefresh
&& !sharedConnection.delegate.equals(failoverTcpConnection.delegate)
&& sharedConnection.isOpen()) {
sharedConnection.close();
}
closeRefreshedIfNecessary(sharedConnection, refreshShared, failoverTcpConnection);
setTheConnection(failoverTcpConnection);
}
return failoverTcpConnection;
}
private void closeRefreshedIfNecessary(FailoverTcpConnection sharedConnection, boolean refreshShared,
FailoverTcpConnection failoverTcpConnection) {
this.creationTime = System.currentTimeMillis();
/*
* We may have simply wrapped the same connection in a new wrapper; don't close.
*/
if (refreshShared && this.closeOnRefresh
&& !sharedConnection.delegate.equals(failoverTcpConnection.delegate)
&& sharedConnection.isOpen()) {
sharedConnection.close();
}
}
@Override
public void start() {
for (AbstractClientConnectionFactory factory : this.factories) {

View File

@@ -536,23 +536,25 @@ The following example shows how to configure a failover client connection factor
NOTE: When using the failover connection factory, the `singleUse` property must be consistent between the factory itself and the list of factories it is configured to use.
The connection factory has two properties when used with a shared connection (`singleUse=false`):
The connection factory has two properties related to failing back, when used with a shared connection (`singleUse=false`):
* `refreshSharedInterval`
* `closeOnRefresh`
These are `0` and `false` to retain the same behavior that existed before the properties were added.
Consider the following scenario based on the above configuration:
Let's say `clientFactory1` cannot establish a connection but `clientFactory2` can.
Each time the `failCF` `getConnection()` method is called, we will again attempt to connect using `clientFactory1`; if successful, the "old" connection will remain open and may be reused in future if the first factory fails once more.
When the `failCF` `getConnection()` method is called after the `refreshSharedInterval` has passed, we will again attempt to connect using `clientFactory1`; if successful, the connection to `clientFactory2` will be closed.
If `closeOnRefresh` is `false`, the "old" connection will remain open and may be reused in future if the first factory fails once more.
Set `refreshSharedInterval` to only attempt to reconnect with the first factory after that time has expired; set it to `Long.MAX_VALUE` if you only want to fail back to the first factory when the current connection fails.
Set `refreshSharedInterval` to only attempt to reconnect with the first factory after that time has expired; setting it to `Long.MAX_VALUE` (default) if you only want to fail back to the first factory when the current connection fails.
Set `closeOnRefresh` to close the "old" connection after a refresh actually creates a new connection.
IMPORTANT: These properties do not apply if any of the delegate factories is a `CachingClientConnectionFactory` because the connection caching is handled there; in that case the list of connection factories will always be consulted to get a connection.
Starting with version 5.3, these default to `Long.MAX_VALUE` and `true` so the factory only attempts to fail back when the current connection fails.
To revert to the default behavior of previous versions, set them to `0` and `false`.
[[tcp-affinity-cf]]
==== TCP Thread Affinity Connection Factory