Details
Major Description
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.
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
Every subsequent call will result in a NoClassDefFoundError, because we already tried to load DynamicClassLoader.class once and failed.
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
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.