[
https://issues.apache.org/jira/browse/HBASE-22701?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Duo Zhang updated HBASE-22701:
------------------------------
Fix Version/s: 3.0.0-alpha-1
> Better handle invalid local directory for DynamicClassLoader
> ------------------------------------------------------------
>
> Key: HBASE-22701
> URL: https://issues.apache.org/jira/browse/HBASE-22701
> Project: HBase
> Issue Type: Improvement
> Components: regionserver
> Reporter: Josh Elser
> Assignee: Josh Elser
> Priority: Major
> Fix For: 3.0.0-alpha-1, 2.3.0, 1.6.0, 2.1.7, 2.2.2, 1.4.12
>
> Attachments: HBASE-22701.001.patch, HBASE-22701.002.patch,
> HBASE-22701.003.patch
>
>
> If you give HBase an {{hbase.local.dir}} (usually, "{{hbase.tmp.dir}}/local")
> which is not writable to it, you will get some weird errors on the scan path.
> I just saw this (again?) with Phoenix.
> Specifically, the first attempt to reference DynamicClassLoader (via
> ProtobufUtil), will result in an ExceptionInInitializationError because the
> unchecked exception coming out of DynamicClassLoader's constructor interrupts
> the loading of {{DynamicClassLoader.class}}.
> {noformat}
> 2019-07-14 06:25:34,284 ERROR
> [RpcServer.Metadata.Fifo.handler=12,queue=0,port=16020]
> coprocessor.MetaDataEndpointImpl: dropTable failed
> org.apache.hadoop.hbase.DoNotRetryIOException:
> java.lang.ExceptionInInitializerError
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.translateException(RpcRetryingCallerImpl.java:221)
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithoutRetries(RpcRetryingCallerImpl.java:194)
> at
> org.apache.hadoop.hbase.client.ScannerCallableWithReplicas$RetryingRPC.call(ScannerCallableWithReplicas.java:387)
> at
> org.apache.hadoop.hbase.client.ScannerCallableWithReplicas$RetryingRPC.call(ScannerCallableWithReplicas.java:361)
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithRetries(RpcRetryingCallerImpl.java:107)
> at
> org.apache.hadoop.hbase.client.ResultBoundedCompletionService$QueueingFuture.run(ResultBoundedCompletionService.java:80)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> Caused by: java.lang.ExceptionInInitializerError
> at
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil.toFilter(ProtobufUtil.java:1598)
> at
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil.toScan(ProtobufUtil.java:1152)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.newRegionScanner(RSRpcServices.java:2967)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.scan(RSRpcServices.java:3301)
> at
> org.apache.hadoop.hbase.client.ScannerCallable.openScanner(ScannerCallable.java:332)
> at
> org.apache.hadoop.hbase.client.ScannerCallable.rpcCall(ScannerCallable.java:242)
> at
> org.apache.hadoop.hbase.client.ScannerCallable.rpcCall(ScannerCallable.java:58)
> at
> org.apache.hadoop.hbase.client.RegionServerCallable.call(RegionServerCallable.java:127)
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithoutRetries(RpcRetryingCallerImpl.java:192)
> ... 7 more
> Caused by: java.lang.RuntimeException: Failed to create local dir
> /hadoopfs/fs1/hbase/local/jars, DynamicClassLoader failed to init
> at
> org.apache.hadoop.hbase.util.DynamicClassLoader.initTempDir(DynamicClassLoader.java:110)
> at
> org.apache.hadoop.hbase.util.DynamicClassLoader.<init>(DynamicClassLoader.java:98)
> at
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil$ClassLoaderHolder.lambda$static$0(ProtobufUtil.java:261)
> at java.security.AccessController.doPrivileged(Native Method)
> at
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil$ClassLoaderHolder.<clinit>(ProtobufUtil.java:260)
> ... 16 more
> {noformat}
> Every subsequent call will result in a NoClassDefFoundError, because we
> already tried to load DynamicClassLoader.class once and failed.
> {noformat}
> 2019-07-14 06:25:34,380 ERROR
> [RpcServer.Metadata.Fifo.handler=2,queue=2,port=16020]
> coprocessor.MetaDataEndpointImpl: dropTable failed
> org.apache.hadoop.hbase.DoNotRetryIOException:
> java.lang.NoClassDefFoundError: Could not initialize class
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil$ClassLoaderHolder
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.translateException(RpcRetryingCallerImpl.java:221)
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithoutRetries(RpcRetryingCallerImpl.java:194)
> at
> org.apache.hadoop.hbase.client.ScannerCallableWithReplicas$RetryingRPC.call(ScannerCallableWithReplicas.java:387)
> at
> org.apache.hadoop.hbase.client.ScannerCallableWithReplicas$RetryingRPC.call(ScannerCallableWithReplicas.java:361)
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithRetries(RpcRetryingCallerImpl.java:107)
> at
> org.apache.hadoop.hbase.client.ResultBoundedCompletionService$QueueingFuture.run(ResultBoundedCompletionService.java:80)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1149)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:624)
> at java.lang.Thread.run(Thread.java:748)
> Caused by: java.lang.NoClassDefFoundError: Could not initialize class
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil$ClassLoaderHolder
> at
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil.toFilter(ProtobufUtil.java:1598)
> at
> org.apache.hadoop.hbase.shaded.protobuf.ProtobufUtil.toScan(ProtobufUtil.java:1152)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.newRegionScanner(RSRpcServices.java:2967)
> at
> org.apache.hadoop.hbase.regionserver.RSRpcServices.scan(RSRpcServices.java:3301)
> at
> org.apache.hadoop.hbase.client.ScannerCallable.openScanner(ScannerCallable.java:332)
> at
> org.apache.hadoop.hbase.client.ScannerCallable.rpcCall(ScannerCallable.java:242)
> at
> org.apache.hadoop.hbase.client.ScannerCallable.rpcCall(ScannerCallable.java:58)
> at
> org.apache.hadoop.hbase.client.RegionServerCallable.call(RegionServerCallable.java:127)
> at
> org.apache.hadoop.hbase.client.RpcRetryingCallerImpl.callWithoutRetries(RpcRetryingCallerImpl.java:192)
> ... 7 more
> {noformat}
> The client gets an error about this, and would presumably know that something
> is amiss, but an operator wouldn't potentially see this on their own.
> I see two options:
> # We abort the RegionServer when the DynamicClassLoader fails to run
> # We catch the exception and treat the DynamicClassLoader as disabled (same
> action as if you had set {{hbase.use.dynamic.jars=false}}).
> I want to do #1 so that we don't propagate bogus configuration, but it feels
> a bit "harsh" to do that. I think #2 is the right solution with a
> big-fat-warning.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
