You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@hbase.apache.org by "Christo Wilson (JIRA)" <ji...@apache.org> on 2010/05/28 00:09:47 UTC
[jira] Updated: (HBASE-2619) HBase shell 'alter' command cannot set
table properties to False
[ https://issues.apache.org/jira/browse/HBASE-2619?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Christo Wilson updated HBASE-2619:
----------------------------------
Description:
The alter table command in the HBase shell is broken. Specifically, attempting to set boolean table options like BLOCKCACHE to False does not result in the expected behavior.
This is due to bugs in the file src/main/ruby/hbase/admin.rb, in the hcd() function. There many statements that look like this:
arg[IN_MEMORY]? JBoolean.valueOf(arg[IN_MEMORY]): HColumnDescriptor::DEFAULT_IN_MEMORY
The intent of this code is to 1) determine if a given parameter is present in the associative array, 2) use the supplied value if it exists, or 3) use the default value if it does not exist. However, in the case of boolean parameters that have been set to False this code instead evaluates the variable (to False) and then chooses the default parameter value, which is incorrect.
The attached patch fixes the issue.
was:
The alter table command in the HBase shell is broken. Specifically, attempting to set boolean table options like BLOCKCACHE to False does not result in the expected behavior.
This is due to bugs in the file src/main/ruby/hbase/admin.rb, in the hcd() function. There many statements that look like this:
arg[IN_MEMORY]? JBoolean.valueOf(arg[IN_MEMORY]): HColumnDescriptor::DEFAULT_IN_MEMORY
The intent of this code is to 1) determine if a given parameter is present in the associative array, 2) use the supplied value if it exists, or 3) use the default value if it does not exist. However, in the case of boolean parameters that have been set to False this code instead evaluates the variable (to False) and then chooses the default parameter value, which is incorrect.
The following patch fixes the issue:
--- hbase-trunk-orig/src/main/ruby/hbase/admin.rb 2010-05-27 14:55:14.000000000 -0700
+++ hbase-trunk/src/main/ruby/hbase/admin.rb 2010-05-27 14:58:26.000000000 -0700
@@ -328,14 +328,14 @@
# TODO: What encoding are Strings in jruby?
return HColumnDescriptor.new(name.to_java_bytes,
# JRuby uses longs for ints. Need to convert. Also constants are String
- arg[VERSIONS]? JInteger.new(arg[VERSIONS]): HColumnDescriptor::DEFAULT_VERSIONS,
- arg[HColumnDescriptor::COMPRESSION]? arg[HColumnDescriptor::COMPRESSION]: HColumnDescriptor::DEFAULT_COMPRESSION,
- arg[IN_MEMORY]? JBoolean.valueOf(arg[IN_MEMORY]): HColumnDescriptor::DEFAULT_IN_MEMORY,
- arg[HColumnDescriptor::BLOCKCACHE]? JBoolean.valueOf(arg[HColumnDescriptor::BLOCKCACHE]): HColumnDescriptor::DEFAULT_BLOCKCACHE,
- arg[HColumnDescriptor::BLOCKSIZE]? JInteger.valueOf(arg[HColumnDescriptor::BLOCKSIZE]): HColumnDescriptor::DEFAULT_BLOCKSIZE,
- arg[HColumnDescriptor::TTL]? JInteger.new(arg[HColumnDescriptor::TTL]): HColumnDescriptor::DEFAULT_TTL,
- arg[HColumnDescriptor::BLOOMFILTER]? arg[HColumnDescriptor::BLOOMFILTER]: HColumnDescriptor::DEFAULT_BLOOMFILTER,
- arg[HColumnDescriptor::REPLICATION_SCOPE]? JInteger.new(arg[REPLICATION_SCOPE]): HColumnDescriptor::DEFAULT_REPLICATION_SCOPE)
+ arg.include?(VERSIONS)? JInteger.new(arg[VERSIONS]): HColumnDescriptor::DEFAULT_VERSIONS,
+ arg.include?(HColumnDescriptor::COMPRESSION)? arg[HColumnDescriptor::COMPRESSION]: HColumnDescriptor::DEFAULT_COMPRESSION,
+ arg.include?(IN_MEMORY)? JBoolean.valueOf(arg[IN_MEMORY]): HColumnDescriptor::DEFAULT_IN_MEMORY,
+ arg.include?(HColumnDescriptor::BLOCKCACHE)? JBoolean.valueOf(arg[HColumnDescriptor::BLOCKCACHE]): HColumnDescriptor::DEFAULT_BLOCKCACHE,
+ arg.include?(HColumnDescriptor::BLOCKSIZE)? JInteger.valueOf(arg[HColumnDescriptor::BLOCKSIZE]): HColumnDescriptor::DEFAULT_BLOCKSIZE,
+ arg.include?(HColumnDescriptor::TTL)? JInteger.new(arg[HColumnDescriptor::TTL]): HColumnDescriptor::DEFAULT_TTL,
+ arg.include?(HColumnDescriptor::BLOOMFILTER)? JBoolean.valueOf(arg[HColumnDescriptor::BLOOMFILTER]): HColumnDescriptor::DEFAULT_BLOOMFILTER,
+ arg.include?(HColumnDescriptor::REPLICATION_SCOPE)? JInteger.new(arg[REPLICATION_SCOPE]): HColumnDescriptor::DEFAULT_REPLICATION_SCOPE)
end
#----------------------------------------------------------------------------------------------
> HBase shell 'alter' command cannot set table properties to False
> ----------------------------------------------------------------
>
> Key: HBASE-2619
> URL: https://issues.apache.org/jira/browse/HBASE-2619
> Project: HBase
> Issue Type: Bug
> Components: scripts
> Affects Versions: 0.20.4
> Environment: CentOS
> Reporter: Christo Wilson
> Priority: Minor
> Fix For: 0.21.0
>
> Original Estimate: 0.02h
> Remaining Estimate: 0.02h
>
> The alter table command in the HBase shell is broken. Specifically, attempting to set boolean table options like BLOCKCACHE to False does not result in the expected behavior.
> This is due to bugs in the file src/main/ruby/hbase/admin.rb, in the hcd() function. There many statements that look like this:
> arg[IN_MEMORY]? JBoolean.valueOf(arg[IN_MEMORY]): HColumnDescriptor::DEFAULT_IN_MEMORY
> The intent of this code is to 1) determine if a given parameter is present in the associative array, 2) use the supplied value if it exists, or 3) use the default value if it does not exist. However, in the case of boolean parameters that have been set to False this code instead evaluates the variable (to False) and then chooses the default parameter value, which is incorrect.
> The attached patch fixes the issue.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.