Ronald Macmaster created HBASE-29377:
----------------------------------------
Summary: ChaosService: Doesn't Support ZK Quorum Strings with Ports
Key: HBASE-29377
URL: https://issues.apache.org/jira/browse/HBASE-29377
Project: HBase
Issue Type: Bug
Components: integration tests
Affects Versions: 2.6.2, 2.5.11
Reporter: Ronald Macmaster
The ChaosAgent doesn't support zookeeper quorum strings which specify custom
ports. example
{code:java}
# hbase.zookeeper.quorum
localhost:2181,localhost:2182,localhost:2183{code}
When starting the chaosagent service, one will receive the following log
messages. The ChaosAgent is setup with a modified zookeeper quorum string
always appending hbase.zookeeper.property.clientPort (2181) to each quorum
server - even though the original quorum prop already specifies a port.
{code:java}
bin/hbase chaosagent -c start
2025-06-04T19:27:39,932 INFO [main] zookeeper.ZooKeeper: Initiating client
connection,
connectString=localhost:2181:2181,localhost:2182:2181,localhost:2183:2181
sessionTimeout=600000 watcher=org.apache.hadoop.hbase.chaos.ChaosAgent@67389cb8
2025-06-04T21:42:38,698 INFO [main-EventThread] chaos.ChaosAgent: Processing
event: WatchedEvent state:Closed type:None path:null
2025-06-04T21:42:38,697 ERROR [main-SendThread()] client.StaticHostProvider:
Unable to resolve address: localhost:2181/<unresolved>:2181
java.net.UnknownHostException: localhost:2181: invalid IPv6 address literal
at
java.net.InetAddress.invalidIPv6LiteralException(InetAddress.java:1693) ~[?:?]
at java.net.InetAddress.getAllByName(InetAddress.java:1663) ~[?:?]
at
org.apache.zookeeper.client.StaticHostProvider$1.getAllByName(StaticHostProvider.java:88)
~[zookeeper-3.8.4.jar:3.8.4]
at
org.apache.zookeeper.client.StaticHostProvider.resolve(StaticHostProvider.java:141)
~[zookeeper-3.8.4.jar:3.8.4]
at
org.apache.zookeeper.client.StaticHostProvider.next(StaticHostProvider.java:368)
~[zookeeper-3.8.4.jar:3.8.4]
at org.apache.zookeeper.ClientCnxn$SendThread.run(ClientCnxn.java:1204)
~[zookeeper-3.8.4.jar:3.8.4]{code}
The behavior comes from the [getZKQuorum() method in the ChaosUtils
class|https://github.com/apache/hbase/blob/d93102ccf631f11028d466acaa1ea3862c185b44/hbase-it/src/main/java/org/apache/hadoop/hbase/chaos/ChaosUtils.java#L36-L43]
which is only used in the [ChaosService
setup|https://github.com/apache/hbase/blob/d93102ccf631f11028d466acaa1ea3862c185b44/hbase-it/src/main/java/org/apache/hadoop/hbase/chaos/ChaosService.java#L79-L80].
This approach deviates from the convention across other areas of the codebase
that instead rely on the [ZKConfig helper
methods|https://github.com/apache/hbase/blob/master/hbase-client/src/main/java/org/apache/hadoop/hbase/zookeeper/ReadOnlyZKClient.java#L136-L141]
to fetch quorum strings.
We should use the the ZKConfig methods instead which are used elsewhere and
[account for the case where the port is already
specified|https://github.com/apache/hbase/blob/d93102ccf631f11028d466acaa1ea3862c185b44/hbase-common/src/main/java/org/apache/hadoop/hbase/zookeeper/ZKConfig.java#L165-L202]
in configuration.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
