This is an automated email from the ASF dual-hosted git repository.
rexxiong pushed a commit to branch branch-0.5
in repository https://gitbox.apache.org/repos/asf/celeborn.git
The following commit(s) were added to refs/heads/branch-0.5 by this push:
new 5a69d1641 [CELEBORN-1524] Support IPv6 hostnames for Apache Ratis
5a69d1641 is described below
commit 5a69d16416d9bace9927486f96fecb8c33705692
Author: Mridul Muralidharan <mridulatgmail.com>
AuthorDate: Fri Jul 26 16:43:09 2024 +0800
[CELEBORN-1524] Support IPv6 hostnames for Apache Ratis
### What changes were proposed in this pull request?
Workaround a Apache Ratis bug in Celeborn until a new Ratis release with
the fix is released which we can use.
### Why are the changes needed?
[RATIS-2131](https://issues.apache.org/jira/browse/RATIS-2131) has been
fixed, and will be available in 3.2.0 - until it is released, this will work
around the issue.
### Does this PR introduce _any_ user-facing change?
Fixes a bug
### How was this patch tested?
Manual testing in IPv6 env with hostnames for ratis config.
Closes #2646 from mridulm/workaround-RATIS-2131.
Authored-by: Mridul Muralidharan <mridulatgmail.com>
Signed-off-by: Shuang <[email protected]>
(cherry picked from commit cc6bc8d18d58b94697e788424442f07f6f743d03)
Signed-off-by: Shuang <[email protected]>
---
.../deploy/master/clustermeta/ha/HARaftServer.java | 24 ++++++++++++++++++++--
.../deploy/master/clustermeta/ha/MasterNode.scala | 6 ------
2 files changed, 22 insertions(+), 8 deletions(-)
diff --git
a/master/src/main/java/org/apache/celeborn/service/deploy/master/clustermeta/ha/HARaftServer.java
b/master/src/main/java/org/apache/celeborn/service/deploy/master/clustermeta/ha/HARaftServer.java
index ff18322fb..b6c12c45a 100644
---
a/master/src/main/java/org/apache/celeborn/service/deploy/master/clustermeta/ha/HARaftServer.java
+++
b/master/src/main/java/org/apache/celeborn/service/deploy/master/clustermeta/ha/HARaftServer.java
@@ -19,6 +19,7 @@ package
org.apache.celeborn.service.deploy.master.clustermeta.ha;
import java.io.File;
import java.io.IOException;
+import java.net.Inet6Address;
import java.net.InetAddress;
import java.net.InetSocketAddress;
import java.nio.charset.StandardCharsets;
@@ -34,6 +35,7 @@ import javax.net.ssl.TrustManager;
import scala.Tuple2;
import com.google.common.annotations.VisibleForTesting;
+import com.google.common.net.InetAddresses;
import com.google.protobuf.InvalidProtocolBufferException;
import org.apache.ratis.RaftConfigKeys;
import org.apache.ratis.client.RaftClientConfigKeys;
@@ -172,6 +174,22 @@ public class HARaftServer {
TimeUnit.MILLISECONDS);
}
+ // copy of org.apache.ratis.util.NetUtils.address2String
+ private static String address2String(InetSocketAddress address) {
+ if (address == null) {
+ return null;
+ }
+ String hostName = address.getHostName();
+ final StringBuilder b = new StringBuilder(hostName);
+ // Surround with '[', ']' only if it is a IPv6 ip - not for a IPv6 host
+ if (address.getAddress() instanceof Inet6Address
+ && InetAddresses.isInetAddress(hostName)
+ && InetAddresses.forString(hostName).getAddress().length == 16) {
+ b.insert(0, '[').append(']');
+ }
+ return b.append(':').append(address.getPort()).toString();
+ }
+
public static HARaftServer newMasterRatisServer(
MetaHandler metaHandler, CelebornConf conf, MasterNode localNode,
List<MasterNode> peerNodes)
throws IOException {
@@ -182,7 +200,9 @@ public class HARaftServer {
RaftPeer localRaftPeer =
RaftPeer.newBuilder()
.setId(localRaftPeerId)
- .setAddress(ratisAddr)
+ // TODO(CELEBORN-1525): Once Ratis 3.2.0 is released and we move
to it, we
+ // should remove address2String and change this to
.setAddress(ratisAddr)
+ .setAddress(address2String(ratisAddr))
.setClientAddress(localNode.rpcEndpoint())
// We use admin address to host the internal rpc address
.setAdminAddress(localNode.internalRpcEndpoint())
@@ -209,7 +229,7 @@ public class HARaftServer {
raftPeer =
RaftPeer.newBuilder()
.setId(raftPeerId)
- .setAddress(peerRatisAddr)
+ .setAddress(address2String(peerRatisAddr))
.setClientAddress(peer.rpcEndpoint())
// We use admin address to host the internal rpc address
.setAdminAddress(peer.internalRpcEndpoint())
diff --git
a/master/src/main/scala/org/apache/celeborn/service/deploy/master/clustermeta/ha/MasterNode.scala
b/master/src/main/scala/org/apache/celeborn/service/deploy/master/clustermeta/ha/MasterNode.scala
index ca4ad8da2..fb3b9b5a2 100644
---
a/master/src/main/scala/org/apache/celeborn/service/deploy/master/clustermeta/ha/MasterNode.scala
+++
b/master/src/main/scala/org/apache/celeborn/service/deploy/master/clustermeta/ha/MasterNode.scala
@@ -41,15 +41,9 @@ case class MasterNode(
def rpcEndpoint: String = rpcHost + ":" + rpcPort
- def rpcIpEndpoint: String = rpcAddr.getAddress.getHostAddress + ":" + rpcPort
-
def internalRpcEndpoint: String = rpcHost + ":" + internalRpcPort
- def internalRpcIpEndpoint: String = rpcAddr.getAddress.getHostAddress + ":"
+ rpcPort
-
lazy val ratisAddr = MasterNode.createSocketAddr(ratisHost, ratisPort)
-
- lazy val rpcAddr = MasterNode.createSocketAddr(rpcHost, rpcPort)
}
object MasterNode extends Logging {
