[ 
https://issues.apache.org/jira/browse/HBASE-19841?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16353436#comment-16353436
 ] 

stack commented on HBASE-19841:
-------------------------------

(Took a while to track down...). The job that fails is the launch of ITBLL 
Generate. IntegrationTestingUtility *subclasses* HTU.  There is then some 
interesting dancing around this ITU with Configuration objects. If an existing 
Configuration, we'll use it when we create an ITU. It will have read in the 
config files and have HDFS as fs set. Along comes this patch and in the 
construction of ITU which calls through to an HTU Constructor, we stamp local 
fs as our fs

351         // Every cluster is a local cluster until we start DFS
352         // Note that conf could be null, but this.conf will not be
353         String dataTestDir = getDataTestDir().toString();
354         this.conf.set("fs.defaultFS","file:///");
355         this.conf.set(HConstants.HBASE_DIR, "file://" + dataTestDir);
356         LOG.debug("Setting {} to {}", HConstants.HBASE_DIR, dataTestDir);
357         
this.conf.setBoolean(CommonFSUtils.UNSAFE_STREAM_CAPABILITY_ENFORCE,false);

The launch gets confused on which fs to use.


> Tests against hadoop3 fail with StreamLacksCapabilityException
> --------------------------------------------------------------
>
>                 Key: HBASE-19841
>                 URL: https://issues.apache.org/jira/browse/HBASE-19841
>             Project: HBase
>          Issue Type: Test
>            Reporter: Ted Yu
>            Assignee: Mike Drob
>            Priority: Major
>             Fix For: 2.0.0-beta-2
>
>         Attachments: 19841.007.patch, 19841.06.patch, 19841.v0.txt, 
> 19841.v1.txt, HBASE-19841.v10.patch, HBASE-19841.v11.patch, 
> HBASE-19841.v11.patch, HBASE-19841.v2.patch, HBASE-19841.v3.patch, 
> HBASE-19841.v4.patch, HBASE-19841.v5.patch, HBASE-19841.v7.patch, 
> HBASE-19841.v8.patch, HBASE-19841.v8.patch, HBASE-19841.v8.patch, 
> HBASE-19841.v9.patch
>
>
> The following can be observed running against hadoop3:
> {code}
> java.io.IOException: cannot get log writer
>       at 
> org.apache.hadoop.hbase.regionserver.TestCompactingMemStore.compactingSetUp(TestCompactingMemStore.java:107)
>       at 
> org.apache.hadoop.hbase.regionserver.TestCompactingMemStore.setUp(TestCompactingMemStore.java:89)
> Caused by: 
> org.apache.hadoop.hbase.util.CommonFSUtils$StreamLacksCapabilityException: 
> hflush and hsync
>       at 
> org.apache.hadoop.hbase.regionserver.TestCompactingMemStore.compactingSetUp(TestCompactingMemStore.java:107)
>       at 
> org.apache.hadoop.hbase.regionserver.TestCompactingMemStore.setUp(TestCompactingMemStore.java:89)
> {code}
> This was due to hbase-server/src/test/resources/hbase-site.xml not being 
> picked up by Configuration object. Among the configs from this file, the 
> value for "hbase.unsafe.stream.capability.enforce" relaxes check for presence 
> of hflush and hsync. Without this config entry,  
> StreamLacksCapabilityException is thrown.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to