https://bugzilla.mcs.anl.gov/swift/show_bug.cgi?id=690 --- Comment #37 from David Kelly <[email protected]> 2012-04-09 23:40:44 --- I tried tonight and can still see this problem using only one machine. Here is how I do this: Run this script to create random files (you could also use dd if=/dev/random... here, but it's slower) --- #!/bin/bash if [ -z "$1" ] || [ -z "$2" ]; then echo "Usage: setup.sh <num_files> <size_in_megabytes>" exit 1 fi NUMFILES=$1 FILESIZE=$2 FILESIZE=$(($FILESIZE*1048576)) for count in `seq 1 $NUMFILES` do FILENAME=`mktemp -p data` openssl rand $FILESIZE -out $FILENAME done --- Run as "./setup.sh 100 200" to create 100 200MB files. The swift script -------------------- type file; app (file o) cat (file i) { cat @i stdout=@o; } file input_files[]<filesys_mapper; location="data", pattern="tmp.*">; foreach j in input_files { file output<single_file_mapper; location="data", file=@strcat(@filename(j), ".out")>; output = cat(j); } -------- Also, be sure to run a swift script after the coaster service starts, but before the worker is started. This is the workaround for bug #467. Without this, the value for jobs per node gets set to 1, and the timeouts don't happen. -- Configure bugmail: https://bugzilla.mcs.anl.gov/swift/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching all bug changes.