[AMBARI-24382] Ambari shouldn't deployed "Undefined" Config Values

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 2.5.2
• Fix Version/s: 2.6.0
• Component/s: ambari-server
• Assignee: Unassigned
• Reporter: Hari Sekhon
• Created: 30/Jul/18 13:36
• Updated: 15/Aug/18 08:40
• Resolved: 15/Aug/18 08:40

Description

Ambari should not deploy any properties with "Undefined" value (breaks HBase Master as shown below).

When using a config group for HBase RegionServers to only enable Bucket Cache on RegionServers (see AMBARI-24370), and then enabling a bunch of settings related to using the bucket cache in the default config, Ambari will infer that there should be a bucketcache setting and injects the following properties with a literal value of "Undefined" in hbase-site.xml:

<property>
<name>hbase.bucketcache.ioengine</name>
<value>Undefined</value>
</property>

<property>
<name>hbase.bucketcache.size</name>
<value>Undefined</value>
</property>

which breaks HMaster restart:

2018-07-30 13:26:08,283 ERROR [main] master.HMasterCommandLine: Master exiting
java.lang.RuntimeException: Failed construction of Master: class org.apache.hadoop.hbase.master.HMaster
at org.apache.hadoop.hbase.master.HMaster.constructMaster(HMaster.java:2824)
at org.apache.hadoop.hbase.master.HMasterCommandLine.startMaster(HMasterCommandLine.java:235)
at org.apache.hadoop.hbase.master.HMasterCommandLine.run(HMasterCommandLine.java:139)
at org.apache.hadoop.util.ToolRunner.run(ToolRunner.java:76)
at org.apache.hadoop.hbase.util.ServerCommandLine.doMain(ServerCommandLine.java:126)
at org.apache.hadoop.hbase.master.HMaster.main(HMaster.java:2838)
Caused by: java.lang.NumberFormatException: For input string: "Undefined"
at sun.misc.FloatingDecimal.readJavaFormatString(FloatingDecimal.java:2043)
at sun.misc.FloatingDecimal.parseFloat(FloatingDecimal.java:122)
at java.lang.Float.parseFloat(Float.java:451)
at org.apache.hadoop.conf.Configuration.getFloat(Configuration.java:1400)
at org.apache.hadoop.hbase.io.hfile.CacheConfig.getBucketCache(CacheConfig.java:597)
at org.apache.hadoop.hbase.io.hfile.CacheConfig.getL2(CacheConfig.java:566)
at org.apache.hadoop.hbase.io.hfile.CacheConfig.instantiateBlockCache(CacheConfig.java:650)
at org.apache.hadoop.hbase.io.hfile.CacheConfig.<init>(CacheConfig.java:239)
at org.apache.hadoop.hbase.regionserver.HRegionServer.<init>(HRegionServer.java:591)
at org.apache.hadoop.hbase.master.HMaster.<init>(HMaster.java:425)
at sun.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
at sun.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:62)
at sun.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
at java.lang.reflect.Constructor.newInstance(Constructor.java:423)
at org.apache.hadoop.hbase.master.HMaster.constructMaster(HMaster.java:2819)
... 5 more

The above two settings are only added to a config group RegionServers, but when the following settings are applied to the default config group thinking they should be ignored by Masters because bucket cache isn't enabled on Master, it turns out that Ambari infers they should exist, leaves them Undefined but deploys them anyway (these settings are from the OpenTSDB HBase performance tuning guide btw):

hbase.rs.cacheblocksonwrite=true
hbase.rs.evictblocksonclose=false
hfile.block.bloom.cacheonwrite=true
hfile.block.index.cacheonwrite=true
hbase.block.data.cachecompressed=true
hbase.bucketcache.blockcache.single.percentage=.99
hbase.bucketcache.blockcache.multi.percentage=0
hbase.bucketcache.blockcache.memory.percentage=.01

I've worked around it by moving the config to the regionservers config group, but ran in to AMBARI-24371 again.

Issue Links

• duplicates: AMBARI-21848 Ambari Replaces the Value of Undefined Service Configuration Properties with "Undefined" in Default Config Group If the Same Property is Defined in Different Config Group
• is related to: AMBARI-24371 HBase RegionServers require stale config restart but Restart RegionServers action shows "no RegionServers to do rolling restarts"!
• is related to: AMBARI-24370 Config Group - unable to remove config setting