Major Description
When we used dynamometer to test HDFS performance, the test encountered a error when generate DataNode Block info, then the generation process failed. The error stack is
Error: java.io.IOException: State transition not allowed; from DEFAULT to FILE_WITH_REPLICATION at com.linkedin.dynamometer.blockgenerator.XMLParser.transitionTo(XMLParser.java:107) at com.linkedin.dynamometer.blockgenerator.XMLParser.parseLine(XMLParser.java:77) at com.linkedin.dynamometer.blockgenerator.XMLParserMapper.map(XMLParserMapper.java:53) at com.linkedin.dynamometer.blockgenerator.XMLParserMapper.map(XMLParserMapper.java:26) at org.apache.hadoop.mapreduce.Mapper.run(Mapper.java:151) at org.apache.hadoop.mapred.MapTask.runNewMapper(MapTask.java:828) at org.apache.hadoop.mapred.MapTask.run(MapTask.java:341) at org.apache.hadoop.mapred.YarnChild$2.run(YarnChild.java:174) at java.security.AccessController.doPrivileged(Native Method) at javax.security.auth.Subject.doAs(Subject.java:415) at org.apache.hadoop.security.UserGroupInformation.doAs(UserGroupInformation.java:1690) at org.apache.hadoop.mapred.YarnChild.main(YarnChild.java:168)
After checking Fsimage xml and the source code, I find that XMLParser can not parse the lines correctly, these lines are like
<directive><id>8963</id><path>/user/somepath/path1</path><replication>3</replication><pool>cache_other_pool</pool><expiration><millis>1544454142310</millis><relatilve>false</relatilve></expiration> <directive><id>8964</id><path>/user/somepath/path2</path><replication>3</replication><pool>cache_hadoop-data_pool</pool><expiration><millis>1544497817686</millis><relatilve>false</relatilve></expiration> <directive><id>8965</id><path>/user/somepath/path3</path><replication>3</replication><pool>cache_hadoop-peisong_pool</pool><expiration><millis>1544451500312</millis><relatilve>false</relatilve></expiration> <directive><id>8967</id><path>/user/somepath/path4</path><replication>3</replication><pool>cache_other_pool</pool><expiration><millis>1544497602570</millis><relatilve>false</relatilve></expiration>
These fsimage xml lines are generated when HDFS Centralized Cache Management (CCM) feature is used.
I have a discuss with xkrogen [here|https://github.com/linkedin/dynamometer/pull/77], and some patches provided can fix the issue.