I recently encountered similar network related errors and was able to fix it by applying the ethtool updates decribed here [ https://issues.apache.org/jira/plugins/servlet/mobile#issue/SPARK-5085]
On Friday, April 29, 2016, Buntu Dev <buntu...@gmail.com> wrote: > Just to provide more details, I have 200 blocks (parquet files) with avg > block size of 70M. When limiting the result set to 100k ("select * from tbl > order by c1 limit 100000") works but when increasing it to say 1M I keep > running into this error: > Connection reset by peer: socket write error > > I would ultimately want to store the result set as parquet. Are there any > other options to handle this? > > Thanks! > > On Wed, Apr 27, 2016 at 11:10 AM, Buntu Dev <buntu...@gmail.com > <javascript:_e(%7B%7D,'cvml','buntu...@gmail.com');>> wrote: > >> I got 14GB of parquet data and when trying to apply order by using spark >> sql and save the first 1M rows but keeps failing with "Connection reset >> by peer: socket write error" on the executors. >> >> I've allocated about 10g to both driver and the executors along with >> setting the maxResultSize to 10g but still fails with the same error. >> I'm using Spark 1.5.1. >> >> Are there any other alternative ways to handle this? >> >> Thanks! >> > >