Hi. I've increased number of JS machines for map and reduce to 32 and now I don't get this strange behaviour (now on every query I get same results count), so it looks like the problem is solved on riak 1.4.10, but I still get different results when I try to use riak 2.1.0 (even if I increase number of JS machines there). Looks like I have to stop using map-reduce in riak 2.1.0
On May 4, 2015, at 7:40 PM, Christopher Meiklejohn <cmeiklej...@basho.com> wrote: > >> On Apr 22, 2015, at 6:45 PM, Sadykov Rustem <sadr...@gmail.com> wrote: >> >> Hello. >> I'm testing riak 1.4.10 on 2 node cluster (on different servers), but result >> is same on 3 nodes cluster and on single node also. >> >> I try to count records in my bucket and always get 3-4 different results. >> Query is: >> >> curl -XPOST http://localhost:8098/mapred -H 'Content-Type: application/json' >> -d '{"inputs":"account", "query":[{"reduce":{"language":"erlang", >> "module":"riak_kv_mapreduce", "function":"reduce_count_inputs", >> "arg":{"reduce_phase_batch_size":1000} } }] }’ > > Hi Sadykov, > > Map/Reduce chooses one replica for each value to execute the query on during > a distributed Map/Reduce job. Given that, subsequent executions might occur > on different replicas given the coverage plan chosen during that period. > > That said, can you provide information on when you are writing the values you > are running the job against? For example, are you writing to a quorum of > nodes right before the query? It’s possible to see some staleness if you are > using majority quorums and you are executing against replicas that are > delayed due to network traffic. > > - Chris > > Christopher Meiklejohn > Senior Software Engineer > Basho Technologies, Inc. > cmeiklej...@basho.com >
_______________________________________________ riak-users mailing list riak-users@lists.basho.com http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
