[
https://issues.apache.org/jira/browse/HIVE-28529?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17883724#comment-17883724
]
Xiaoqiao He edited comment on HIVE-28529 at 9/23/24 7:21 AM:
-------------------------------------------------------------
{quote}In hadoop maybe we can explore having a method for cloning without lock,
but if someone concurrently modifies the config passed while creating the
HMSHandler, ideally not, but if yes, that would backfire
{quote}
+1. I plan originally to update hadoop Configuration implement at first glance,
however I also think it will not be safe way. Then try to reach out here and
look forwards to some more smooth solution. Thanks all.
was (Author: hexiaoqiao):
{code:java}
In hadoop maybe we can explore having a method for cloning without lock, but if
someone concurrently modifies the config passed while creating the HMSHandler,
ideally not, but if yes, that would backfire
{code}
+1. I plan originally to update hadoop Configuration implement at first glance,
however I also think it will not be safe way. Then try to reach out here and
look forwards to some more smooth solution. Thanks all.
> HiveMetaStore#getConf blocked when meet high load
> -------------------------------------------------
>
> Key: HIVE-28529
> URL: https://issues.apache.org/jira/browse/HIVE-28529
> Project: Hive
> Issue Type: Improvement
> Security Level: Public(Viewable by anyone)
> Components: Metastore
> Reporter: Xiaoqiao He
> Priority: Major
>
> Thousand threads are blocked for long time when metastore meet high load as
> the following stack shows.
> a. there are 1836 threads(as stack 1 shows) are waiting Lock
> #0x00007f8bf9477180 which is hold by stack 2.
> {code:java}
> # grep "0x00007f8bf9477180" metastore.stack | wc -l
> 1836
> {code}
> b. there are 105 threads (as stack 2 shows) are waiting Lock
> #0x00007f8bf805f660 which is hold by stack 3.
> {code:java}
> # grep "0x00007f8bf805f660" metastore.stack | wc -l
> 105
> {code}
> c. stack 3 shows that it is time cost operation when init configuration,
> which hold object (#hiveConf as the last code snippet)synchronized which is
> at key path for metastore and impact the performance.
> So, IMO, it need to improve and remove the lock competition to improve the
> performance. FYI.
> NOTE: I have deployed one early version, but the newest one include this
> issue too.
> {code:java}
> "pool-12-thread-1482355" #126195588 prio=5 os_prio=0 tid=0x00007f86d507b800
> nid=0x21570 waiting for monitor entry [0x00007f875849b000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at
> org.apache.hadoop.hive.metastore.MetaStoreInit.initConnectionUrlHook(MetaStoreInit.java:95)
> - waiting to lock <0x00007f8bf9477180> (a java.lang.Class for
> org.apache.hadoop.hive.metastore.MetaStoreInit)
> at
> org.apache.hadoop.hive.metastore.MetaStoreInit.updateConnectionURL(MetaStoreInit.java:62)
> at
> org.apache.hadoop.hive.metastore.RawStoreProxy.init(RawStoreProxy.java:87)
> at
> org.apache.hadoop.hive.metastore.RawStoreProxy.<init>(RawStoreProxy.java:55)
> at
> org.apache.hadoop.hive.metastore.RawStoreProxy.getProxy(RawStoreProxy.java:66)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.newRawStore(HiveMetaStore.java:817)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.getMS(HiveMetaStore.java:795)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.get_database_core(HiveMetaStore.java:1308)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.get_database(HiveMetaStore.java:1240)
> at sun.reflect.GeneratedMethodAccessor59.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.hadoop.hive.metastore.RetryingHMSHandler.invoke(RetryingHMSHandler.java:276)
> at com.sun.proxy.$Proxy23.get_database(Unknown Source)
> at sun.reflect.GeneratedMethodAccessor59.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$MetricHMSProxy.invoke(HiveMetaStore.java:8241)
> at com.sun.proxy.$Proxy23.get_database(Unknown Source)
> at
> org.apache.hadoop.hive.metastore.api.ThriftHiveMetastore$Processor$get_database.getResult(ThriftHiveMetastore.java:11142)
> at
> org.apache.hadoop.hive.metastore.api.ThriftHiveMetastore$Processor$get_database.getResult(ThriftHiveMetastore.java:11126)
> at org.apache.thrift.ProcessFunction.process(ProcessFunction.java:39)
> at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:39)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor$1.run(HadoopThriftAuthBridge.java:754)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor$1.run(HadoopThriftAuthBridge.java:749)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1717)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor.process(HadoopThriftAuthBridge.java:749)
> at
> org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:286)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> {code}
> {code:java}
> "pool-12-thread-1482175" #126193367 prio=5 os_prio=0 tid=0x00007f87567d3000
> nid=0x20565 waiting for monitor entry [0x00007f8698ccd000]
> java.lang.Thread.State: BLOCKED (on object monitor)
> at
> org.apache.hadoop.conf.Configuration.getProps(Configuration.java:2415)
> - waiting to lock <0x00007f8bf805f660> (a
> org.apache.hadoop.hive.conf.HiveConf)
> at org.apache.hadoop.conf.Configuration.get(Configuration.java:1235)
> at
> org.apache.hadoop.hive.metastore.MetaStoreInit.initConnectionUrlHook(MetaStoreInit.java:96)
> - locked <0x00007f8bf9477180> (a java.lang.Class for
> org.apache.hadoop.hive.metastore.MetaStoreInit)
> at
> org.apache.hadoop.hive.metastore.MetaStoreInit.updateConnectionURL(MetaStoreInit.java:62)
> at
> org.apache.hadoop.hive.metastore.RawStoreProxy.init(RawStoreProxy.java:87)
> at
> org.apache.hadoop.hive.metastore.RawStoreProxy.<init>(RawStoreProxy.java:55)
> at
> org.apache.hadoop.hive.metastore.RawStoreProxy.getProxy(RawStoreProxy.java:66)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.newRawStore(HiveMetaStore.java:817)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.getMS(HiveMetaStore.java:795)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.get_table_by_view(HiveMetaStore.java:2247)
> at sun.reflect.GeneratedMethodAccessor100.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.hadoop.hive.metastore.RetryingHMSHandler.invoke(RetryingHMSHandler.java:276)
> at com.sun.proxy.$Proxy23.get_table_by_view(Unknown Source)
> at sun.reflect.GeneratedMethodAccessor100.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$MetricHMSProxy.invoke(HiveMetaStore.java:8241)
> at com.sun.proxy.$Proxy23.get_table_by_view(Unknown Source)
> at
> org.apache.hadoop.hive.metastore.api.ThriftHiveMetastore$Processor$get_table_by_view.getResult(ThriftHiveMetastore.java:11812)
> at
> org.apache.hadoop.hive.metastore.api.ThriftHiveMetastore$Processor$get_table_by_view.getResult(ThriftHiveMetastore.java:11796)
> at org.apache.thrift.ProcessFunction.process(ProcessFunction.java:39)
> at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:39)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor$1.run(HadoopThriftAuthBridge.java:754)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor$1.run(HadoopThriftAuthBridge.java:749)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1717)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor.process(HadoopThriftAuthBridge.java:749)
> at
> org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:286)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> {code}
> {code:java}
> "pool-12-thread-1477440" #126119273 prio=5 os_prio=0 tid=0x00007f86d53fb800
> nid=0xa5d runnable [0x00007f86fe5a3000]
> java.lang.Thread.State: RUNNABLE
> at
> java.util.concurrent.ConcurrentHashMap.tryPresize(ConcurrentHashMap.java:2334)
> at
> java.util.concurrent.ConcurrentHashMap.putAll(ConcurrentHashMap.java:1082)
> at
> java.util.concurrent.ConcurrentHashMap.<init>(ConcurrentHashMap.java:852)
> at org.apache.hadoop.conf.Configuration.<init>(Configuration.java:716)
> - locked <0x00007f8bf805f660> (a org.apache.hadoop.hive.conf.HiveConf)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.getConf(HiveMetaStore.java:723)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$HMSHandler.setMetaConf(HiveMetaStore.java:744)
> at sun.reflect.GeneratedMethodAccessor39.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.hadoop.hive.metastore.RetryingHMSHandler.invoke(RetryingHMSHandler.java:276)
> at com.sun.proxy.$Proxy23.setMetaConf(Unknown Source)
> at sun.reflect.GeneratedMethodAccessor39.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> at
> org.apache.hadoop.hive.metastore.HiveMetaStore$MetricHMSProxy.invoke(HiveMetaStore.java:8241)
> at com.sun.proxy.$Proxy23.setMetaConf(Unknown Source)
> at
> org.apache.hadoop.hive.metastore.api.ThriftHiveMetastore$Processor$setMetaConf.getResult(ThriftHiveMetastore.java:11090)
> at
> org.apache.hadoop.hive.metastore.api.ThriftHiveMetastore$Processor$setMetaConf.getResult(ThriftHiveMetastore.java:11074)
> at org.apache.thrift.ProcessFunction.process(ProcessFunction.java:39)
> at org.apache.thrift.TBaseProcessor.process(TBaseProcessor.java:39)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor$1.run(HadoopThriftAuthBridge.java:754)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor$1.run(HadoopThriftAuthBridge.java:749)
> at java.security.AccessController.doPrivileged(Native Method)
> at javax.security.auth.Subject.doAs(Subject.java:422)
> at
> org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1717)
> at
> org.apache.hadoop.hive.thrift.HadoopThriftAuthBridge$Server$TUGIAssumingProcessor.process(HadoopThriftAuthBridge.java:749)
> at
> org.apache.thrift.server.TThreadPoolServer$WorkerProcess.run(TThreadPoolServer.java:286)
> at
> java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1142)
> at
> java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:617)
> at java.lang.Thread.run(Thread.java:745)
> {code}
> org.apache.hadoop.hive.metastore.HiveMetaStore.HMSHandler#getConf
> {code:java}
> @Override
> public Configuration getConf() {
> Configuration conf = threadLocalConf.get();
> if (conf == null) {
> conf = new Configuration(hiveConf);
> threadLocalConf.set(conf);
> }
> return conf;
> }
> {code}
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
