accumulo-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Christopher Tubbs" <ctubbsii...@apache.org>
Subject Re: Review Request 23413: ACCUMULO-2988 Upgrade from 1.4 direct to 1.6
Date Sat, 12 Jul 2014 03:05:33 GMT

-----------------------------------------------------------
This is an automatically generated e-mail. To reply, visit:
https://reviews.apache.org/r/23413/#review47681
-----------------------------------------------------------


Thanks for doing this! Would be great to see this get in for 1.6.1. Pushing ACCUMULO-2145
might make it easier to test this.


server/base/src/main/java/org/apache/accumulo/server/Accumulo.java
<https://reviews.apache.org/r/23413/#comment83816>

    I'd think this method could just be dropped in favor of the new one. Is there any reason
we'd still call this?



server/base/src/main/java/org/apache/accumulo/server/Accumulo.java
<https://reviews.apache.org/r/23413/#comment83817>

    This is kind of an important check. Perhaps move this to a method like "boolean canUpgradeFromDataVersion(int
version);" with a comment?



server/base/src/main/java/org/apache/accumulo/server/ServerConstants.java
<https://reviews.apache.org/r/23413/#comment83818>

    We have many different kinds of versions. Please clarify the constant (something like
"TWO_DATA_VERSIONS_AGO"), so it's not misused.



server/master/src/main/java/org/apache/accumulo/master/Master.java
<https://reviews.apache.org/r/23413/#comment83819>

    prefer putting this check into a method (canUpgrade? needsUpgrade?... whatever is most
appropriate)



server/master/src/main/java/org/apache/accumulo/master/Master.java
<https://reviews.apache.org/r/23413/#comment83820>

    should be in a descriptive function (see previous comments)



server/master/src/main/java/org/apache/accumulo/master/Master.java
<https://reviews.apache.org/r/23413/#comment83821>

    Avoid the string "!METADATA". Log messages should refer to a generic "metadata table"
or similar, to be consistent with previous changes, which purged specific table names from
being referenced (especially when it's a string literal instead of a constant). This is for
future-proofing any other metadata changes we might make later.


- Christopher Tubbs


On July 11, 2014, 3:19 p.m., Sean Busbey wrote:
> 
> -----------------------------------------------------------
> This is an automatically generated e-mail. To reply, visit:
> https://reviews.apache.org/r/23413/
> -----------------------------------------------------------
> 
> (Updated July 11, 2014, 3:19 p.m.)
> 
> 
> Review request for accumulo.
> 
> 
> Bugs: ACCUMULO-2988
>     https://issues.apache.org/jira/browse/ACCUMULO-2988
> 
> 
> Repository: accumulo
> 
> 
> Description
> -------
> 
> forward ports of some patches around 1.4 -> 1.5 upgrades to the 1.6 branch, as well
as new additions.
> 
> 
> Diffs
> -----
> 
>   README c2c2446dc0312c64427c2d7205056c72b2cfdec2 
>   server/base/src/main/java/org/apache/accumulo/server/Accumulo.java 68d862d5baaad04ff38789155c0fb60ac18f92b8

>   server/base/src/main/java/org/apache/accumulo/server/ServerConstants.java b577abbb72c9dce70b8f12dfefe381d38c48cb8e

>   server/base/src/main/java/org/apache/accumulo/server/util/MetadataTableUtil.java 374017d6544a3be5bccb70eab7a994a92f9693ac

>   server/master/src/main/java/org/apache/accumulo/master/Master.java c367ae0c2febbb297924da4a87ef1d91104d94c4

> 
> Diff: https://reviews.apache.org/r/23413/diff/
> 
> 
> Testing
> -------
> 
> ran through test upgrade
> 
> * Install 1.4.5
> * Load a variety of table storage configs, each wi/156M cells (~4GB uncompressed)
> ** no compression, bloom filter on, block cache on
> ** lzo compression, bloom filter off, block cache off
> ** snappy compression, bloom filters on, block cache off
> ** snappy compression, bloom filters off, block cache on
> * Upgrade according to README to 1.6.1-SNAP+patches
> * Verify correct contents readable
> * force compaction
> * restart cluster
> * Verify correct contents readable
> 
> 
> Thanks,
> 
> Sean Busbey
> 
>


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