accumulo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From ke...@deenlo.com
Subject Re: Review Request 24862: Accumulo-3021 Limit size of split rows
Date Wed, 20 Aug 2014 19:11:59 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/24862/#review51121
-----------------------------------------------------------



server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java
<https://reviews.apache.org/r/24862/#comment89151>

    whats this log warn for?



server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java
<https://reviews.apache.org/r/24862/#comment89152>

    I think it would be ok to remove this check and let it complain about the big row.



test/src/test/java/org/apache/accumulo/test/LargeSplitRowIT.java
<https://reviews.apache.org/r/24862/#comment89145>

    Can remove this comment.   Since tserver config is being used, can not use shared tserver
instance.



test/src/test/java/org/apache/accumulo/test/LargeSplitRowIT.java
<https://reviews.apache.org/r/24862/#comment89153>

    May need to sleep longer here because the thread that checks for splits on the tserver
only runs once every second.   
    
    Alternatively, could decrease config for TSERV_MAJC_DELAY from 1s to 100ms or 50ms.  That
would probably be better.



test/src/test/java/org/apache/accumulo/test/LargeSplitRowIT.java
<https://reviews.apache.org/r/24862/#comment89150>

    Instead of doing a sleep then assert, could do something like the following.
    
    while(conn.tableOperations().listSplits(tableName).size() == 0){
      sleep(250)
    }
    
    This approach will allow the test to pass on slower machines.  Eventually the entire rest
will fail with a timeout if the condition is not met.
    
    If there are other places where the # of splits should change, then can use this approach.
 Situation where the # of splits should stay the same is tricker, not sure what we could do
besides sleep and check in that case.


- kturner


On Aug. 20, 2014, 6:28 p.m., Jenna Huston wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/24862/
> -----------------------------------------------------------
> 
> (Updated Aug. 20, 2014, 6:28 p.m.)
> 
> 
> Review request for accumulo and kturner.
> 
> 
> Bugs: Accumulo-3021
>     https://issues.apache.org/jira/browse/Accumulo-3021
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> Added constraint for maximum end row size. Checked this constraint before any split is
done.
> A unit test was added to test the added constraint when both a user initiaed a tablet
split and when an automatic split occurs.  This test also tests to make sure that when new
data is added to a table that is small enough to be used as an end row the table splits.
> 
> 
> Diffs
> -----
> 
>   core/src/main/java/org/apache/accumulo/core/conf/Property.java 85c56f8 
>   server/tserver/src/main/java/org/apache/accumulo/tserver/tablet/Tablet.java 82045b6

>   test/src/test/java/org/apache/accumulo/test/LargeSplitRowIT.java PRE-CREATION 
> 
> Diff: https://reviews.apache.org/r/24862/diff/
> 
> 
> Testing
> -------
> 
> Passes unit tests.
> 
> 
> Thanks,
> 
> Jenna Huston
> 
>


Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message