Re: Why broadcast Edge property?

2015-03-26 Thread Jianfeng (Jeff) Zhang
OrderedPartitionedKVEdgeConfig is only for ScatterGather. UnorderedKVEdgeConfig is for broadcast and 1-1 edge. Usually broadcast output is not sorted OrderedPartitionedKVEdgeConfig.java public EdgeProperty createDefaultEdgeProperty() { EdgeProperty edgeProperty = EdgeProperty.create

Re: Why broadcast Edge property?

2015-03-26 Thread Azuryy Yu
Thanks Jianfeng. if I use OrderedPartitionedKVEdgeConfig , and use broadcast edge property with parallism 1, does that Tez also do shuffle and group before broadcast? On Fri, Mar 27, 2015 at 12:47 PM, Jianfeng (Jeff) Zhang < jzh...@hortonworks.com> wrote: > > Hi Azuryy, > > I check the sourc

Re: Why broadcast Edge property?

2015-03-26 Thread Jianfeng (Jeff) Zhang
Hi Azuryy, I check the source code, the last vertex Writer do an in-memory sort in its processor. So in this case broadcast is possible (also require parallelism to be 1) If the edge is scatter-gather and use OrderedPartitionedKVEdgeConfig, then the in-memory sort is not necessary because the

Why broadcast Edge property?

2015-03-26 Thread Azuryy Yu
Hi, please look through this simple code: https://github.com/sequenceiq/sequenceiq-samples/blob/master/tez-topk/src/main/java/com/sequenceiq/tez/topk/TopK.java why they create a broadcast edge property from SUM to WRITER? what about default edge property? (scatter-gather)

Re: Tez blog on Hortonworks is outdate

2015-03-26 Thread Azuryy Yu
Yes. Hitesh, I mainly refer to the Tez docs, just found this blog deprecated occasionaly. On Thu, Mar 26, 2015 at 11:41 PM, Hitesh Shah wrote: > Thanks for the heads up, Azuryy. I will pass the information along to the > relevant folks at Hortonworks. And yes, you are right. At some point, we

Re: Tez blog on Hortonworks is outdate

2015-03-26 Thread Hitesh Shah
Thanks for the heads up, Azuryy. I will pass the information along to the relevant folks at Hortonworks. And yes, you are right. At some point, we merged the code into the common TezClient so that it became easier for the user to toggle session-mode on/off via configuration and not have to write

Re: BufferTooSmallException

2015-03-26 Thread Cyrille Chépélov
Hi, I'm the original victim :) just sent up TEZ-2237. Sent as much logs as was practical up to this point; can supply on a direct basis as much as required to nail the issue. To give some context: these two failing DAG are part of a meta-DAG comprised of 20 distinct DAG, all generated throug

Tez blog on Hortonworks is outdate

2015-03-26 Thread Azuryy Yu
Hi, Please review this blog, http://www.hortonworks.com/blog/introducing-tez-sessions/ espcially for the section Using Tez Sessions TezSession is removed, right? if we want to re-use session in Tez, just do like this: TezConfiguration tezConf = new TezConfiguration(); tezConf.setBoolean(TezConfi