[
https://issues.apache.org/jira/browse/ATLAS-1408?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15765891#comment-15765891
]
Hemanth Yamijala commented on ATLAS-1408:
-----------------------------------------
[~madhan.neethiraj] pointed me to the '.' being the problem. Irrespective of
the solutions to this specific problem, it would be nice to not get the server
into an unrecoverable state.
> Creating an attribute definition of array type, where the elements have a dot
> in the typename causes issues
> -----------------------------------------------------------------------------------------------------------
>
> Key: ATLAS-1408
> URL: https://issues.apache.org/jira/browse/ATLAS-1408
> Project: Atlas
> Issue Type: Bug
> Affects Versions: 0.8-incubating
> Reporter: Hemanth Yamijala
> Priority: Critical
>
> I created a type File.v1 and then another type FileSet.v1 which contains an
> array of File.v1 type instances. The creation was using the
> {{AtlasClient.createType(TypeDef)}} API. This returned back successfully.
> However, post this Atlas server went into a funny state where simple DSL
> queries like {{hive_table}} started failing. Looking the application logs,
> noticed this error post the registration of the FileSet.v1 type:
> {code}
> 2016-12-20 13:46:53,960 INFO - [pool-2-thread-4 -
> 8eaf688d-9cd4-46dc-92da-d84991512e9b:] ~ TypeSystem reset invoked by
> TypeRegistry changes (DefaultMetadataService:749)
> .... Many types restored and then ....
> 2016-12-20 13:46:54,266 ERROR - [pool-2-thread-4 -
> 8eaf688d-9cd4-46dc-92da-d84991512e9b:] ~ Failed to restore type-system after
> TypeRegistry changes (DefaultMetadataService:759)
> org.apache.atlas.typesystem.exception.TypeNotFoundException: Unknown
> datatype: array<HdfsFile.v1>
> at
> org.apache.atlas.typesystem.types.TypeSystem.getDataType(TypeSystem.java:192)
> at
> org.apache.atlas.typesystem.types.TypeSystem$TransientTypeSystem.getDataType(TypeSystem.java:669)
> at
> org.apache.atlas.typesystem.types.AttributeInfo.<init>(AttributeInfo.java:46)
> at
> org.apache.atlas.typesystem.types.TypeSystem$TransientTypeSystem.constructAttributeInfo(TypeSystem.java:496)
> at
> org.apache.atlas.typesystem.types.TypeSystem$TransientTypeSystem.constructHierarchicalType(TypeSystem.java:540)
> at
> org.apache.atlas.typesystem.types.TypeSystem$TransientTypeSystem.orderAndConstructTypes(TypeSystem.java:583)
> at
> org.apache.atlas.typesystem.types.TypeSystem$TransientTypeSystem.verifyTypes(TypeSystem.java:709)
> at
> org.apache.atlas.typesystem.types.TypeSystem.createTransientTypeSystem(TypeSystem.java:346)
> at
> org.apache.atlas.services.DefaultMetadataService.onChange(DefaultMetadataService.java:753)
> at
> org.apache.atlas.repository.store.graph.AtlasTypeDefGraphStore$TypeRegistryUpdateHook.notifyListeners(AtlasTypeDefGraphStore.java:940)
> at
> org.apache.atlas.repository.store.graph.AtlasTypeDefGraphStore$TypeRegistryUpdateHook.onComplete(AtlasTypeDefGraphStore.java:924)
> {code}
> DSL failures post this error had this trace:
> {code}
> 2016-12-20 13:49:55,398 ERROR - [pool-2-thread-8 -
> 3d671edf-9dce-4656-a5e3-04e3d3bd3ade:] ~ Unable to get entity list for
> dslQuery hive_table (MetadataDiscoveryResource:141)
> org.apache.atlas.discovery.DiscoveryException: Invalid expression : hive_table
> at
> org.apache.atlas.discovery.graph.GraphBackedDiscoveryService.evaluate(GraphBackedDiscoveryService.java:139)
> at
> org.apache.atlas.discovery.graph.GraphBackedDiscoveryService.searchByDSL(GraphBackedDiscoveryService.java:124)
> at
> org.apache.atlas.GraphTransactionInterceptor.invoke(GraphTransactionInterceptor.java:51)
> at
> org.apache.atlas.web.resources.MetadataDiscoveryResource.searchUsingQueryDSL(MetadataDiscoveryResource.java:135)
> at sun.reflect.GeneratedMethodAccessor41.invoke(Unknown Source)
> at
> sun.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
> at java.lang.reflect.Method.invoke(Method.java:498)
> {code}
> Restarting the server caused more issues because the types restoration failed
> as in the error in the first stack trace, which caused failures in
> initializations and so, effectively, the server did not come up. The
> workaround was the delete the HBase titan table and start over again.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
