[
https://issues.apache.org/jira/browse/HBASE-29654?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Daniel Roudnitsky updated HBASE-29654:
--------------------------------------
Description:
+Problem+
BinaryComponentComparator filter enables a user to compare against a subset of
a byte array by specifying an offset into the byte array from which the
comparison should start. The issue here is that if the offset provided by the
user is longer than a byte array encountered by the scan query (e.g an offset
of 40 is used, and there is a "short" byte array of length 38 somewhere in the
table), when the scan query reaches the "short" byte array the query will fail
in a nongraceful manner - the scan RPC fails with a mysterious
ArrayIndexOutOfBoundsException exception, the client will continue to retry the
failed RPC excessively until exhausting all retries, and the client operation
ultimately fail out with an unintuitive RetriesExhaustedException.
+Root cause+
Serverside the scan request handler will throw an unhandled/unexpected
ArrayIndexOutOfBoundsException during request processing when applying the
filter and attempting to do an unchecked byte array comparison using an offset
which extends beyond the length of an encountered byte array, when this happens
the client will treat the remote exception from the server as an IOException
(which wraps the remote exception), and because IOException is retryable, the
client will proceed to exhaust all of its retries re-running the same query,
all of which are guaranteed to fail, and will eventually fail in a
nonobvious/nongraceful manner with a stack trace that looks like:
{code:java}
org.apache.hadoop.hbase.client.RetriesExhaustedException: Failed after
attempts=2, exceptions:
2025-09-26T19:09:35.531Z,
RpcRetryingCaller{globalStartTime=2025-09-26T19:09:35.412Z, pause=1000,
maxAttempts=2}, java.io.IOException: java.io.IOException
at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:451)
at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
Caused by: java.lang.ArrayIndexOutOfBoundsException
2025-09-26T19:09:36.659Z,
RpcRetryingCaller{globalStartTime=2025-09-26T19:09:35.412Z, pause=1000,
maxAttempts=2}, java.io.IOException: java.io.IOException
at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:451)
at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
Caused by: java.lang.ArrayIndexOutOfBoundsException
at
org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithRetries(RpcRetryingCallerImpl.java:142)
at
org.apache.hadoop.hbase.client.ResultBoundedCompletionService$QueueingFuture.run(ResultBoundedCompletionService.java:73)
at
java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
at
java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
at java.base/java.lang.Thread.run(Thread.java:840){code}
+Proposed solution+
Instead of the server throwing an unhandled/unexpected
ArrayIndexOutOfBoundsException from BinaryComponentComparator, it gracefully
return a DoNotRetryIOException which will prevent the client from attempting
excessive retries which are guaranteed to fail, and the DoNotRetryIOException
should clearly explain the root cause to the user - the user provided byte
offset for the filter exceeded the length of a byte array that the scan
encountered.
was:
+Problem+
BinaryComponentComparator filter enables a user to compare against a subset of
a byte array by specifying an offset into the byte array from which the
comparison should start. The issue here is that if the offset provided by the
user is longer than a byte array encountered by the scan query (e.g an offset
of 40 is used, and there is a "short" byte array of length 38 somewhere in the
table), when the scan query reaches the short byte array the query will fail in
a nongraceful manner - the scan RPC fails with a mysterious remote
ArrayIndexOutOfBoundsException exception, the client will continue to retry the
failed RPC excessively until exhausting all retries, and the client ultimately
fails with an unintuitive RetriesExhaustedException.
+Root cause+
Serverside the scan request handler will throw an unhandled/unexpected
ArrayIndexOutOfBoundsException during request processing when applying the
filter and attempting to do an unchecked byte array comparison using an offset
which extends beyond the length of an encountered byte array, when this happens
the client will treat the remote exception from the server as an IOException
(which wraps the remote exception), and because IOException is retryable, the
client will proceed to exhaust all of its retries re-running the same query,
all of which are guaranteed to fail, and will eventually fail in a nonobvious
and nongraceful manner with a stack trace that looks like:
{code:java}
org.apache.hadoop.hbase.client.RetriesExhaustedException: Failed after
attempts=2, exceptions:
2025-09-26T19:09:35.531Z,
RpcRetryingCaller{globalStartTime=2025-09-26T19:09:35.412Z, pause=1000,
maxAttempts=2}, java.io.IOException: java.io.IOException
at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:451)
at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
Caused by: java.lang.ArrayIndexOutOfBoundsException
2025-09-26T19:09:36.659Z,
RpcRetryingCaller{globalStartTime=2025-09-26T19:09:35.412Z, pause=1000,
maxAttempts=2}, java.io.IOException: java.io.IOException
at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:451)
at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369)
at
org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
Caused by: java.lang.ArrayIndexOutOfBoundsException
at
org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithRetries(RpcRetryingCallerImpl.java:142)
at
org.apache.hadoop.hbase.client.ResultBoundedCompletionService$QueueingFuture.run(ResultBoundedCompletionService.java:73)
at
java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
at
java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
at java.base/java.lang.Thread.run(Thread.java:840){code}
+Proposed solution+
Instead of the server throwing an unhandled/unexpected
ArrayIndexOutOfBoundsException from BinaryComponentComparator, it gracefully
return a DoNotRetryIOException which will prevent the client from attempting
excessive retries which are guaranteed to fail, and the DoNotRetryIOException
should clearly explain the root cause to the user - the user provided byte
offset for the filter exceeded the length of a byte array that the scan
encountered.
> BinaryComponentComparator filter fails non gracefully with
> ArrayIndexOutOfBoundsException
> -----------------------------------------------------------------------------------------
>
> Key: HBASE-29654
> URL: https://issues.apache.org/jira/browse/HBASE-29654
> Project: HBase
> Issue Type: Bug
> Components: Filters
> Reporter: Daniel Roudnitsky
> Assignee: Daniel Roudnitsky
> Priority: Major
>
> +Problem+
> BinaryComponentComparator filter enables a user to compare against a subset
> of a byte array by specifying an offset into the byte array from which the
> comparison should start. The issue here is that if the offset provided by the
> user is longer than a byte array encountered by the scan query (e.g an offset
> of 40 is used, and there is a "short" byte array of length 38 somewhere in
> the table), when the scan query reaches the "short" byte array the query will
> fail in a nongraceful manner - the scan RPC fails with a mysterious
> ArrayIndexOutOfBoundsException exception, the client will continue to retry
> the failed RPC excessively until exhausting all retries, and the client
> operation ultimately fail out with an unintuitive RetriesExhaustedException.
> +Root cause+
> Serverside the scan request handler will throw an unhandled/unexpected
> ArrayIndexOutOfBoundsException during request processing when applying the
> filter and attempting to do an unchecked byte array comparison using an
> offset which extends beyond the length of an encountered byte array, when
> this happens the client will treat the remote exception from the server as an
> IOException (which wraps the remote exception), and because IOException is
> retryable, the client will proceed to exhaust all of its retries re-running
> the same query, all of which are guaranteed to fail, and will eventually fail
> in a nonobvious/nongraceful manner with a stack trace that looks like:
> {code:java}
> org.apache.hadoop.hbase.client.RetriesExhaustedException: Failed after
> attempts=2, exceptions:
> 2025-09-26T19:09:35.531Z,
> RpcRetryingCaller{globalStartTime=2025-09-26T19:09:35.412Z, pause=1000,
> maxAttempts=2}, java.io.IOException: java.io.IOException
> at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:451)
> at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
> Caused by: java.lang.ArrayIndexOutOfBoundsException
> 2025-09-26T19:09:36.659Z,
> RpcRetryingCaller{globalStartTime=2025-09-26T19:09:35.412Z, pause=1000,
> maxAttempts=2}, java.io.IOException: java.io.IOException
> at org.apache.hadoop.hbase.ipc.RpcServer.call(RpcServer.java:451)
> at org.apache.hadoop.hbase.ipc.CallRunner.run(CallRunner.java:139)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:369)
> at
> org.apache.hadoop.hbase.ipc.RpcExecutor$Handler.run(RpcExecutor.java:349)
> Caused by: java.lang.ArrayIndexOutOfBoundsException
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithRetries(RpcRetryingCallerImpl.java:142)
> at
> org.apache.hadoop.hbase.client.ResultBoundedCompletionService$QueueingFuture.run(ResultBoundedCompletionService.java:73)
> at
> java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1136)
> at
> java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:635)
> at java.base/java.lang.Thread.run(Thread.java:840){code}
> +Proposed solution+
> Instead of the server throwing an unhandled/unexpected
> ArrayIndexOutOfBoundsException from BinaryComponentComparator, it gracefully
> return a DoNotRetryIOException which will prevent the client from attempting
> excessive retries which are guaranteed to fail, and the DoNotRetryIOException
> should clearly explain the root cause to the user - the user provided byte
> offset for the filter exceeded the length of a byte array that the scan
> encountered.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
