Details
-
Bug
-
Status: Resolved
-
Major
-
Resolution: Invalid
-
3.0.1
-
None
-
None
-
Ambari 2.7.1
HDP 3.0.1
Isilon OneFS 8.1.2
Description
Background on ats-hbase start success:: We had setStoragePolicy exception on OneFS for yarn-ats hbase region server and below messages respectively on HDP and OneFS cluster.
HDP Host è /var/log/hadoop-yarn/embedded-yarn-ats-hbase/
2018-11-02 05:07:21,027 WARN [StoreOpener-1588230740-1] util.CommonFSUtils: Unable to set storagePolicy=HOT for path=hdfs://kb-hdp-1.west.isilon.com:8020/atsv2/hbase/data/data/hbase/meta/1588230740/info. DEBUG log level might have more details.
java.lang.reflect.InvocationTargetException
at sun.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
OneFS HDFS log for (unsecured)
2018-11-01T10:25:19Z <30.6> kbhusan-2oq1zs7-1 hdfs[2604]: [hdfs] RPC V9
user: yarn-ats exception: org.apache.hadoop.ipc.RpcNoSuchMethodException
cause: Unknown RPC: setStoragePolicy
OneFS HDFS log for (secured)
2018-11-01T12:39:56Z <30.6> kbhusan-2oq1zs7-1 hdfs[2604]: [hdfs] RPC V9
user: yarn-ats-hbase/krb-hdp301.solarch.emc.com@SOLARCH.EMC.COM exception:
org.apache.hadoop.ipc.RpcNoSuchMethodException cause: Unknown RPC:
setStoragePolicy
As a resolution, we set below 2 properties in custom hbase-site.xml file and setstoragePolicy exception got resolved.
Custom yarn-hbase-site.xml
'hbase.wal.storage.policy' to ‘NONE’
‘hbase.hstore.block.storage.policy’ to ‘NONE’
*********
NOW, we have a new issue that is OneFS HDFS log and /var/log/hadoop-yarn/embedded-yarn-ats-hbase/hbase-yarn-ats-regionserver-***.log gets flooded with below error messages.*OneFS HDFS Log
-------------------
2018-11-06T06:25:43Z <30.6> kbhusan-edl9tdc-1 hdfs[2542]: [hdfs] RPC V9
complete user:
yarn-ats-hbase/rduvnode343393.west.isilon.com@WEST.ISILON.COM exception:
java.io.FileNotFoundException cause: File not found:
/atsv2/hbase/data/WALs/rduvnode343393.west.isilon.com,17020,1541478673646/[rduvnode343393.west.isilon.com|http://rduvnode343393.west.isilon.com/]%2C17020%2C1541478673646.meta.1541478709060.meta
<FS Exception at CreateHandleFromPath():448 status:
STATUS_OBJECT_NAME_NOT_FOUND = 0xC0000034 Path:
atsv2/hbase/data/WALs/rduvnode343393.west.isilon.com,17020,1541478673646/[rduvnode343393.west.isilon.com|http://rduvnode343393.west.isilon.com/]%2C17020%2C1541478673646.meta.1541478709060.meta>
---------------
/var/log/hadoop-yarn/embedded-yarn-ats-hbase/hbase-yarn-ats-regionserver-**.log
--------------
2018-11-06 04:01:12,863 INFO [LruBlockCacheStatsExecutor]
hfile.LruBlockCache: totalSize=1.12 MB, freeSize=1.25 GB, max=1.25 GB,
blockCount=24, accesses=287, hits=263, hitRatio=91.64%, ,
cachingAccesses=287, cachingHits=263, cachingHitsRatio=91.64%,
evictions=419, evicted=0, evictedPerRun=0.0
2018-11-06 04:01:48,763 WARN
LeaseRenewer:[yarn-ats@kb-hdp301-2.west.isilon.com impl.LeaseRenewer:
Failed to renew lease for [DFSClient_NONMAPREDUCE_-533598397_1] for 30
seconds. Will retry shortly ...
org.apache.hadoop.ipc.RemoteException(org.apache.hadoop.security.authorize.AuthorizationException):
Client auth method 'SIMPLE', mechanism '' is not supported on zone:1
auth:'kerberos_only'
at org.apache.hadoop.ipc.Client.getRpcResponse(Client.java:1497)
at org.apache.hadoop.ipc.Client.call(Client.java:1443)
at org.apache.hadoop.ipc.Client.call(Client.java:1353)
“
“
“
“
2018-11-06 12:51:02,843 WARN [Close-WAL-Writer-1] asyncfs.FanOutOneBlockAsyncDFSOutputHelper: complete file /atsv2/hbase/data/WALs/rduvnode343393.west.isilon.com,17020,1541488822956/[rduvnode343393.west.isilon.com|http://rduvnode343393.west.isilon.com/]%2C17020%2C1541488822956.1541492429802 failed, retry = 637
org.apache.hadoop.ipc.RemoteException(java.io.FileNotFoundException): File not found: /atsv2/hbase/data/WALs/rduvnode343393.west.isilon.com,17020,1541488822956/[rduvnode343393.west.isilon.com|http://rduvnode343393.west.isilon.com/]%2C17020%2C1541488822956.1541492429802
at org.apache.hadoop.ipc.Client.getRpcResponse(Client.java:1497)
at org.apache.hadoop.ipc.Client.call(Client.java:1443)
at org.apache.hadoop.ipc.Client.call(Client.java:1353)
--------------
On checking the ats-hbase app logs, found hbase region server is trying to flush the old edits to free the WAL and it fails due to authentication issue.
Why yarn-ats hbase expects OneFS(namenode and datanode) to have a yarn principle?