You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@accumulo.apache.org by Chris McCubbin <ch...@sqrrl.com> on 2013/11/25 22:37:05 UTC

Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/
-----------------------------------------------------------

(Updated Nov. 25, 2013, 9:37 p.m.)


Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.


Changes
-------

added accumulo group


Repository: accumulo


Description
-------

This is a prototype implementation of ACCUMULO-1787.


Diffs
-----

  core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
  server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 

Diff: https://reviews.apache.org/r/15846/diff/


Testing
-------


Thanks,

Chris McCubbin


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by Chris McCubbin <ch...@sqrrl.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/
-----------------------------------------------------------

(Updated Nov. 24, 2014, 1:31 p.m.)


Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.


Changes
-------

Link to the JIRA issue.


Bugs: ACCUMULO-1787
    https://issues.apache.org/jira/browse/ACCUMULO-1787


Repository: accumulo


Description
-------

This is a prototype implementation of ACCUMULO-1787.


Diffs
-----

  core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
  server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 

Diff: https://reviews.apache.org/r/15846/diff/


Testing
-------


Thanks,

Chris McCubbin


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by Chris McCubbin <ch...@sqrrl.com>.

> On Nov. 26, 2013, 2:32 p.m., kturner wrote:
> > server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java, line 300
> > <https://reviews.apache.org/r/15846/diff/2/?file=391121#file391121line300>
> >
> >     The following situation is kinda bugging me, but I do not think its very important.
> >     
> >     1. some snappy files exceed threshold 
> >     2. creates a gzip file thats below threshold
> >     3. forced compaction of that gzip file creates a snappy file
> >     
> >     
> >     
> >

I had actually thought about this as a potential issue as well. Can it actually happen though? Can the gzip file below the threshold be compacted again without combining with another file?


- Chris


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/#review29433
-----------------------------------------------------------


On Nov. 26, 2013, 4:32 a.m., Chris McCubbin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15846/
> -----------------------------------------------------------
> 
> (Updated Nov. 26, 2013, 4:32 a.m.)
> 
> 
> Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This is a prototype implementation of ACCUMULO-1787.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
>   server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 
> 
> Diff: https://reviews.apache.org/r/15846/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chris McCubbin
> 
>


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by Josh Elser <jo...@gmail.com>.

> On Nov. 26, 2013, 2:32 p.m., kturner wrote:
> > server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java, line 296
> > <https://reviews.apache.org/r/15846/diff/2/?file=391121#file391121line296>
> >
> >     Bulk imported files may not have an estimated size.  I can not remember, something to look into.

I believe you're correct, Keith.


- Josh


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/#review29433
-----------------------------------------------------------


On Nov. 26, 2013, 4:32 a.m., Chris McCubbin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15846/
> -----------------------------------------------------------
> 
> (Updated Nov. 26, 2013, 4:32 a.m.)
> 
> 
> Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This is a prototype implementation of ACCUMULO-1787.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
>   server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 
> 
> Diff: https://reviews.apache.org/r/15846/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chris McCubbin
> 
>


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by ke...@deenlo.com.

> On Nov. 26, 2013, 2:32 p.m., kturner wrote:
> > server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java, line 300
> > <https://reviews.apache.org/r/15846/diff/2/?file=391121#file391121line300>
> >
> >     The following situation is kinda bugging me, but I do not think its very important.
> >     
> >     1. some snappy files exceed threshold 
> >     2. creates a gzip file thats below threshold
> >     3. forced compaction of that gzip file creates a snappy file
> >     
> >     
> >     
> >
> 
> Chris McCubbin wrote:
>     I had actually thought about this as a potential issue as well. Can it actually happen though? Can the gzip file below the threshold be compacted again without combining with another file?

A user initiated compaction will compact all files in a tablet, even if there is only one file.   This is useful for applying filters, etc.


- kturner


-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/#review29433
-----------------------------------------------------------


On Nov. 26, 2013, 4:32 a.m., Chris McCubbin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15846/
> -----------------------------------------------------------
> 
> (Updated Nov. 26, 2013, 4:32 a.m.)
> 
> 
> Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This is a prototype implementation of ACCUMULO-1787.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
>   server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 
> 
> Diff: https://reviews.apache.org/r/15846/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chris McCubbin
> 
>


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by ke...@deenlo.com.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/#review29433
-----------------------------------------------------------



server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java
<https://reviews.apache.org/r/15846/#comment56649>

    Bulk imported files may not have an estimated size.  I can not remember, something to look into.



server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java
<https://reviews.apache.org/r/15846/#comment56650>

    The following situation is kinda bugging me, but I do not think its very important.
    
    1. some snappy files exceed threshold 
    2. creates a gzip file thats below threshold
    3. forced compaction of that gzip file creates a snappy file
    
    
    
    


- kturner


On Nov. 26, 2013, 4:32 a.m., Chris McCubbin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15846/
> -----------------------------------------------------------
> 
> (Updated Nov. 26, 2013, 4:32 a.m.)
> 
> 
> Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This is a prototype implementation of ACCUMULO-1787.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
>   server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 
> 
> Diff: https://reviews.apache.org/r/15846/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chris McCubbin
> 
>


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by Chris McCubbin <ch...@sqrrl.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/
-----------------------------------------------------------

(Updated Nov. 26, 2013, 4:32 a.m.)


Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.


Changes
-------

Patch updated with suggestions. Moved to a small file threshold so the default behavior is the same as before.


Repository: accumulo


Description
-------

This is a prototype implementation of ACCUMULO-1787.


Diffs (updated)
-----

  core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
  server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 

Diff: https://reviews.apache.org/r/15846/diff/


Testing
-------


Thanks,

Chris McCubbin


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by Adam Fuchs <af...@apache.org>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/#review29395
-----------------------------------------------------------



core/src/main/java/org/apache/accumulo/core/conf/Property.java
<https://reviews.apache.org/r/15846/#comment56597>

    Whitespace changes should be removed.



core/src/main/java/org/apache/accumulo/core/conf/Property.java
<https://reviews.apache.org/r/15846/#comment56599>

    Instead of specifying the compression threshold as a count in "millions of bytes" it should use PropertyType.MEMORY.



core/src/main/java/org/apache/accumulo/core/conf/Property.java
<https://reviews.apache.org/r/15846/#comment56603>

    We need a way to determine whether this option has been enabled, and configuration should default to the previous behavior if this is not enabled. This could be done with a file compression threshold of 0B, and then the alternative compression type would be for small files instead of large files.



server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java
<https://reviews.apache.org/r/15846/#comment56606>

    This should only be done if the alternative compression type is enabled. If we default the threshold to 0B and flip the > to a < that would be sufficient.


- Adam Fuchs


On Nov. 25, 2013, 9:37 p.m., Chris McCubbin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15846/
> -----------------------------------------------------------
> 
> (Updated Nov. 25, 2013, 9:37 p.m.)
> 
> 
> Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This is a prototype implementation of ACCUMULO-1787.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
>   server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 
> 
> Diff: https://reviews.apache.org/r/15846/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chris McCubbin
> 
>


Re: Review Request 15846: Prototype implementation of ACCUMULO-1787.

Posted by Eric Newton <er...@gmail.com>.
-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/15846/#review29396
-----------------------------------------------------------



core/src/main/java/org/apache/accumulo/core/conf/Property.java
<https://reviews.apache.org/r/15846/#comment56598>

    Use a default of "50M"... it will be consistent with other size properties.  Should use the PropertyType.MEMORY.



server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java
<https://reviews.apache.org/r/15846/#comment56600>

    Just use acuTableConf.getMemoryInBytes, which will handle magical "M" and "G" suffixes.


- Eric Newton


On Nov. 25, 2013, 9:37 p.m., Chris McCubbin wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/15846/
> -----------------------------------------------------------
> 
> (Updated Nov. 25, 2013, 9:37 p.m.)
> 
> 
> Review request for accumulo, Adam Fuchs, Josh Elser, and kturner.
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> This is a prototype implementation of ACCUMULO-1787.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java f4c1778 
>   server/src/main/java/org/apache/accumulo/server/tabletserver/Compactor.java 8e4af64 
> 
> Diff: https://reviews.apache.org/r/15846/diff/
> 
> 
> Testing
> -------
> 
> 
> Thanks,
> 
> Chris McCubbin
> 
>