DESCRIPTION

CBQE-246: add limits to employeeDataSet

Change-Id: Ib97ca71c58dc92e98fc877bd33b5721cceb3fa24


COMMENTS

author: Trevor Santana
date: 2012-06-18 17:07:06.862000000

Uploaded patch set 2.

-------------------------------------
author: Hugo Blankenship
date: 2012-06-18 17:07:10.351000000

Patch Set 2:

Build Started http://qa.hq.northscale.net/job/testrunner-Jasmin Rangel-master/997/ 

-------------------------------------
author: Titus Ho
date: 2012-06-18 17:14:03.712000000

Patch Set 2: Looks good to me, approved

Hey Farshid, we loop even if result is greater/less than expected, but will quit if the result is the same for "max-dupe-result-count' times , which is parameterized and currently set to 10.

tc.input.param('max-dupe-result-count', 10)

-------------------------------------
author: Hugo Blankenship
date: 2012-06-18 17:16:32.708000000

Patch Set 2: Verified

Build Successful 

http://qa.hq.northscale.net/job/testrunner-Jasmin Rangel-master/997/ : SUCCESS

-------------------------------------
author: Zackery Barr
date: 2012-06-18 17:24:56.161000000

Patch Set 2:

can we change the default value to 3 if # of docs is < 50k and have it as 10 otherwise ?

-------------------------------------
author: Titus Ho
date: 2012-06-18 17:58:03.222000000

Patch Set 2:

>can we change the default value to 3 if # of docs is < 50k and have it as 10 otherwise ?

I think it's best to parameterize as needed for certain tests.  

Reason being that dupe_count isn't related to how many documents are in the bucket but rather how efficient the indexer is given a certain test and how many results are expected.   i.e say you rebalance going on with 5 parallel queries attempting to index entire db,  you will get more dupe results as indexes may build slower.  

I'm not opposed to using something other than 10 as default if we know what optimal value is here bc 20mins is lots o time...

-------------------------------------
