@Aaron I believe the issues Josh is talking about are related to
Specifically I think there were issues with using multiple accumulo
instances and multiple sets of authorizations because the input splits
would be initialized with connection information from the configuration
object when being consumed by a task attempt, versus having that
information baked in at creation time on client. This caused issues with
Pig when it would launch MapReduce jobs that used multiple input sources.
On Thu, Dec 19, 2013 at 7:52 AM, Aaron <[EMAIL PROTECTED]> wrote:
> We are currently using the accumulo-pig 1.5 branch (not the 1783-1.5
> branch..that did not work) against Accumulo 1.5 and we don't "appear" to be
> having any issues. I was curious what "issue" exists in order to check our
> On Thu, Dec 19, 2013 at 12:31 AM, Josh Elser <[EMAIL PROTECTED]> wrote:
>> There isn't currently a fully-working version against 1.5.0 due to an
>> issue in Accumulo's InputFormat.
>> There is, however, a working version against 1.5.1-SNAPSHOT if you don't
>> have a problem building Accumulo. Take a look at the ACCUMULO-1783-1.5
>> branch in . A walkthrough on some of the things you can do can be found
>> at .
>> FWIW, I'm also working to get this included in Pig-0.13  which will
>> completely remove the necessity for the accumulo-pig project.
>> Things will hopefully sync up with "real" releases once we can get
>> Accumulo 1.5.1 out the door.
>> - Josh
>>  https://git-wip-us.apache.org/repos/asf?p=accumulo-pig.git
>>  http://people.apache.org/~elserj/accumulo-pig/
>>  https://issues.apache.org/jira/browse/PIG-3573
>> On 12/18/13, 11:56 PM, A.Khanolkar wrote:
>>> Do any of you have a stable version of Accumulo-Pig working with
>>> Accumulo 1.5?
>>> - Anagha Khanolkar