benchwarmer

#!/bin/bash

source ./config
source ./params

# Take clients and rate limit from command line if passed, use defaults
# otherwise

# Clients
CLIENTS=4
if [ -n "$1" ]; then
  CLIENTS="$1"
fi

# Rate limit
if [ -n "$2" ]; then
  RATE="$2"
  RATEOPT="-R $2 "
else
  RATE="NULL"
  RATEOPT=""
fi

# Client limit
CLIENT_LIMIT=0
if [ -n "$3" ]; then
  CLIENT_LIMIT="$3"
fi

# latency limit
if [ -n "$4" ]; then
  LAT="$4"
  LATOPT="-L $4 "
else
  LAT="NULL"
  LATOPT=""
fi

function detect_cpus {
  # getconf should work on Linux and more recent Macs.  Try the "online" processor
  # count first.
  if [ -z "$CPUS" ]; then
    CPUS=`getconf _NPROCESSORS_ONLN 2>/dev/null`
  fi

  # This will work on older Macs and possibly other BSD systems, but is expected
  # to fail on Linux.
  CPUS=`sysctl -n hw.ncpu 2>/dev/null`

  # Newer Macs dropped ncpu sysctl to break out logical and physical cores
  if [ -z "$CPUS" ]; then
    CPUS=`sysctl -n hw.logicalcpu 2>/dev/null`
  fi

  # Try the gnu coreutils getconf, available on newer Linux and Homebrew.
  if [ -z "$CPUS" ]; then
    CPUS=`nproc 2>/dev/null`
  fi

  # Fallback to the getconf total number if it's not available.  That's not
  # always what we want, but we shouldn't even reach this point.
  if [ -z "$CPUS" ]; then
    CPUS=`getconf _NPROCESSORS_CONF 2>/dev/null`
  fi

  # TODO Test whether any of these CPU counters work on FreeBSD and Solaris

  if [ -z "$CPUS" ]; then
    # If there's no more accurate guess, running make with 4 processes works
    # well enough on typical hardware circa 2020.  
    CPUS=4
  fi
}

# Query the test database and put its background writer statistics into
# a set of environment variables, probably for use inserting stats into
# the results database.
#
# This presumes that you have already set $BACKEND_FSYNC before
# calling.
function get_bgwriter {
  BGW=`$TESTPSQL -A -t -F" " -c "select checkpoints_timed,checkpoints_req,buffers_checkpoint,buffers_clean,maxwritten_clean,buffers_backend,buffers_alloc,$BACKEND_FSYNC from pg_stat_bgwriter"`
  set -- $BGW
  checkpoints_timed=$1
  checkpoints_req=$2
  buffers_checkpoint=$3
  buffers_clean=$4
  maxwritten_clean=$5
  buffers_backend=$6
  buffers_alloc=$7
  buffers_backend_fsync=$8
}

# Figure out how many transactions per client, then recompute
# a new total; this copes with rounding issues
if [ -z "$TOTTRANS" ] ; then
  TOTTRANS=1000
fi
TRANS=`expr $TOTTRANS / $CLIENTS`
TOTTRANS=`expr $TRANS \* $CLIENTS`

# If a runtime has been specified instead, throw away the transaction counts
# use it instead
if [ -n "$RUNTIME" ] ; then
  TESTLEN="-T $RUNTIME"
  TOTTRANS=""
else
  TESTLEN="-t $TRANS"
fi

if [ -z "$TRIM" ] ; then
  TRIM=0
fi

# Pass through prepared statements option
PREPAREDOPT=""
if [ -z "$PREPARED" ] ; then
  PREPARED=0
fi
if [ "$PREPARED" -gt 0 ] ; then
    PREPAREDOPT="-M prepared"
fi

# Set MAX_WORKERS automatically on Linux, OS X, and maybe other platforms
if [ "$MAX_WORKERS" == "auto" ] ; then
  detect_cpus
  if [ -z "$CPUS" ] ; then
    error 16 "Bug in detect_cpus function, returned a null value"
  fi
  MAX_WORKERS="$CPUS"
  echo Workers using $MAX_WORKERS CPUs
fi

# Set WORKERS string so that the largest possible worker count
# up to MAX_WORKERS is used, while still dividing CLIENTS into an
# even number per worker.
WORKERS=""
NUM_WORKERS="1"

if [ -n "$MAX_WORKERS" ] ; then

  # Only bother with/allow adjustment to WORKERS if the max is >1.
  # That keeps up out of trouble if using a pgbench before 9.0,
  # where using any value for "-j" won't be allowed, as long as the
  # config file we're given isn't setup incorrectly.

  if [ "$MAX_WORKERS" -gt 1 ]; then
    NUM_WORKERS=$MAX_WORKERS

    while [ "$NUM_WORKERS" -gt 1 ]; do
      (( remainder=$CLIENTS % $NUM_WORKERS ))
      if [ $remainder -eq 0 ] ; then
        break
      fi
      (( NUM_WORKERS = $NUM_WORKERS - 1 ))
    done

    WORKERS="-j ${NUM_WORKERS}"
  fi
fi

# Setting SOCKETS to 1 avoids host IP networking to prefer a socket connection.
# Only useful when running this script on the server itself for local testing.
# TODO Allow specifying the location of the sockets file.
TESTHOSTPARAM="-h $TESTHOST"
if [ -n "$SOCKETS" ]; then
  if [ "$SOCKETS" -eq "1" ]; then
    TESTHOSTPARAM=""
  fi
fi

# psql statements for the test database and the result database
TESTPSQL="psql $TESTHOSTPARAM -U $TESTUSER -p $TESTPORT -d $TESTDB"
RESULTPSQL="psql -h $RESULTHOST -U $RESULTUSER -p $RESULTPORT -d $RESULTDB"
echo Running tests using:  $TESTPSQL
echo Storing results using:  $RESULTPSQL

# See if this database has all the standard pgbench tables in it
PGCOUNT=`$TESTPSQL -A -t -c "SELECT count(*)  FROM pg_stat_user_tables WHERE relname IN ('pgbench_history','pgbench_tellers','pgbench_accounts','pgbench_branches')"`
if [ "$PGCOUNT" -eq 4 ] ; then
  TABLEPREFIX="pgbench_"
  PGBENCH_TABLES=1
  echo Found standard pgbench tables with prefix=$TABLEPREFIX
else
  TABLEPREFIX=""
  PGCOUNT=`$TESTPSQL -A -t -c "SELECT count(*)  FROM pg_stat_user_tables WHERE relname IN ('history','tellers','accounts','branches')"`
  if [ "$PGCOUNT" -eq 4 ] ; then
    echo Found standard pgbench tables with prefix=$TABLEPREFIX
    PGBENCH_TABLES=1
  else
    echo Did not find standard pgbench tables
    PGBENCH_TABLES=0
  fi
fi

# Determine database scale.  Allow overriding it in the config file.
# That's needed to run the pgbench init script.  It can't be an empty string.
if [ -z "$SCALE" ] ; then
  if [ "$PGBENCH_TABLES" -eq "1" ] ; then
    SCALE=`$TESTPSQL -A -t -c "select count(*) from ${TABLEPREFIX}branches"`
  fi
fi
if [ -z "$SCALE" ] ; then
  SCALE="0"
fi

# Confirm we have a useful pgbench to run
if [ -z $PGBENCHBIN ] || [ ! -x $PGBENCHBIN ]; then
  echo ERROR:  cannot find pgbench binary $PGBENCHBIN , aborting
  exit
fi

# Find current test set
SET=`$RESULTPSQL -A -t -c "select max(set) from testset"`
if [ "$?" -ne "0" ]; then
  echo ERROR:  Attempt to determine test set failed
  # TODO Create a fake first set if this happens?  Right now,
  # the likely case is that the test DDL was never added, which
  # makes that sort of correction attempt unlikely to be useful
  exit
fi

# Version of Server
SERVER_VERSION=`$TESTPSQL -A -t -F" " -c "select current_setting('server_version_num')::numeric;"`
VERSION_TEXT=`$TESTPSQL -Atc "select version();"`

# Cleanup pgbench tables, unless we've been told to skip that
if [ "$SKIPINIT" -ne "1" ]; then
  echo Cleaning up database $TESTDB
  if [ "$PGBENCH_TABLES" -eq "1" ] ; then
    $TESTPSQL -c "truncate table ${TABLEPREFIX}history"
  fi

  # If the previous test ran recently and it did a cleanup, we should
  # be safe to skip this possibly lengthy step.  "Recently" is defined
  # as within 10 seconds of when the last test cleanup finished.  There
  # is some margin for error here, particulary if processing the results
  # latency log was very time consuming.  The price of making a mistake
  # is just some extra run-time though, doing a redundant vacuum.  On
  # databases sizes where the vacuum time is very long, its actually less
  # likely that the results log is large.  Tests against large database
  # tend to run slowly.
  CLEANED=`$RESULTPSQL -A -t -c "SELECT EXISTS(SELECT 1 FROM tests WHERE server='${SERVERNAME}' AND cleanup IS NOT NULL AND ((now() - end_time) + cleanup) < '100 seconds'::interval ORDER BY test DESC LIMIT 1)"`
  if [ "$CLEANED" = 'f' ] ; then
    $TESTPSQL -c "vacuum analyze"
  else
    echo "Skipping vacuum, it was recently ran by the last test"
  fi
fi

# Run a reset script if one is provided.  Typical use is cache clearing.
if [ -n "$TESTRESET" ]; then
  $TESTRESET $SERVER_VERSION
fi

# Try to start the test close to the beginning of the checkpoint cycle.
if [ "$SKIPINIT" -ne "1" ]; then
  # We want clean stats from the pg_stat_bgwriter, but those won't show up
  # until after the checkpoint is done.  Wait a bit for the stats to update
  STARTCHECK=`$TESTPSQL -At -c "SELECT checkpoints_req FROM pg_stat_bgwriter"`
  $TESTPSQL -c "checkpoint"
  echo "Waiting for checkpoint statistics"
  while [ 1 ] ; do
    CHECK=`$TESTPSQL -At -c "SELECT checkpoints_req FROM pg_stat_bgwriter"`
    if [ "$CHECK" -gt "$STARTCHECK" ] ; then
      break
    fi 
    sleep 1
  done
fi

# Create the tests record
DBSIZE=`$TESTPSQL -A -t -c "select pg_database_size(current_database())"`
$RESULTPSQL -q -c "insert into tests (server,server_version,script,clients,workers,set,scale,dbsize,rate_limit,client_limit,multi) values('$SERVERNAME','$VERSION_TEXT','$SCRIPT','$CLIENTS','$NUM_WORKERS','$SET','$SCALE','$DBSIZE',$RATE, $CLIENT_LIMIT,$MULTI)"
TEST=`$RESULTPSQL -A -t -c "select max(test) from tests WHERE server='${SERVERNAME}'"`
if [ "$?" -ne "0" ]; then
  echo ERROR  Can\'t read from tests table.  Was the test data installed?
  exit
fi

if [ -z "$TEST" ]; then
  echo ERROR:  Attempt to get a test number returned \"$TEST\", aborting
  exit
fi

# Figure out if this version of PostgreSQL includes buffers_backend_fsync
$TESTPSQL -c "SELECT buffers_backend_fsync FROM pg_stat_bgwriter" >> /dev/null 2>&1
if [ "$?" -ne "0" ]; then
  BACKEND_FSYNC="0"
else
  BACKEND_FSYNC="buffers_backend_fsync"
fi

# Grab starting values for statistics
get_bgwriter
if [ "$SERVER_VERSION" -gt "100000" ]; then
START_WAL=`$TESTPSQL -A -t -F" " -c "select pg_current_wal_lsn()"`
else
START_WAL=`$TESTPSQL -A -t -F" " -c "select pg_current_xlog_location()"`
fi
$RESULTPSQL -c "insert into test_bgwriter(server,test,checkpoints_timed,checkpoints_req,buffers_checkpoint,buffers_clean,maxwritten_clean,buffers_backend,buffers_alloc,buffers_backend_fsync) values('$SERVERNAME','$TEST','$checkpoints_timed','$checkpoints_req','$buffers_checkpoint','$buffers_clean','$maxwritten_clean','$buffers_backend','$buffers_alloc','$buffers_backend_fsync')"
$TESTPSQL -c 'SELECT pg_stat_reset()' >> /dev/null 2>&1
$TESTPSQL -c 'SELECT pg_stat_statements_reset()' >> /dev/null 2>&1

# Tack on multiplier to rate option set
if [[ "$MULTI" -ge "0" ]] ; then
  RATEOPT="$RATEOPT -Dmultiplier=$MULTI"
fi

# Decide between regular script and pass through.
# Most options are lost in pass-through mode to support the init workload.
# pgbench doesn't allow most of its command line parameters unless you're in
# "benchmark mode", which an init operation is not.
PREFIX="${SCRIPT:0:1}"
if [ "$PREFIX" == ":" ] ; then
  SCRIPTOPT="${SCRIPT:1}"
  PASSTHRU=1
else
  SCRIPTOPT="-f $BASEDIR/$TESTDIR/$SCRIPT.sql $TESTLEN -n -l -c $CLIENTS $WORKERS $RATEOPT $LATOPT $PREPAREDOPT"
  PASSTHRU=0
fi

# Setup test directory tree
echo This is test $TEST
mkdir -p results/$SERVERNAME/$TEST

# Start background daemon collectors
OS=`uname`
if [ "$OSDATA" -eq "1" ]; then
  ./timed-os-stats vmstat > results/$SERVERNAME/$TEST/vmstat.log &
  VMSTAT=$!
  ./timed-os-stats iostat > results/$SERVERNAME/$TEST/iostat.log &
  IOSTAT=$!
  if [ "$OS" = "Linux" ] ; then
    ./timed-os-stats meminfo > results/$SERVERNAME/$TEST/meminfo.log &
    MEMINFO=$!
  fi
fi

# This one should run even if osdata collectors are not available
./timed-os-stats $TESTPSQL -At -f $BASEDIR/util/metric-query.sql > results/$SERVERNAME/$TEST/psql.log &
PSQLMET=$!

# Run the main pgbench test
cd results/$SERVERNAME/$TEST
echo Script $SCRIPT executing for $CLIENTS concurrent users... 1>&2

$PGBENCHBIN $SCRIPTOPT -s $SCALE $TESTHOSTPARAM -p $TESTPORT -U $TESTUSER $TESTDB > results.txt 2>&1 &
P=$!
wait $P
$RESULTPSQL -q -c "UPDATE tests SET end_time=now() WHERE server='${SERVERNAME}' AND test='${TEST}'"
$RESULTPSQL -q -c "UPDATE tests SET start_latency=start_time, end_latency=end_time WHERE server='${SERVERNAME}' AND tests.test='$TEST';"

../../../kill_pg "${PSQLMET}"
if [ "$OSDATA" -eq "1" ]; then
  # The PIDs we have will be the process ID of the Python timed-os-stats
  # process, but we must make sure to also kill the child processes (launched
  # via subprocess.Popen()) so they are not lingering forever.
  ../../../kill_pg "${VMSTAT}"
  ../../../kill_pg "${IOSTAT}"
  if [ "$OS" = "Linux" ] ; then
    ../../../kill_pg "${MEMINFO}"
    # Find largest dirty memory value (which is in kB)
    max_dirty=`cat meminfo.log | grep Dirty | awk '{print $4}' | sort -n | tail -n 1`
  fi
fi

# max_dirty has to be a valid integer value
if [ -z "$max_dirty" ] ; then
 max_dirty=0 
fi

# Update bgwriter data with delta
get_bgwriter
$RESULTPSQL -c "update test_bgwriter set \
  checkpoints_timed=$checkpoints_timed - checkpoints_timed,\
  checkpoints_req=$checkpoints_req - checkpoints_req,\
  buffers_checkpoint=$buffers_checkpoint - buffers_checkpoint,\
  buffers_clean=$buffers_clean - buffers_clean,\
  maxwritten_clean=$maxwritten_clean - maxwritten_clean,\
  buffers_backend=$buffers_backend - buffers_backend,\
  buffers_alloc=$buffers_alloc - buffers_alloc,\
  buffers_backend_fsync=$buffers_backend_fsync - buffers_backend_fsync, \
  max_dirty=1024::bigint * '$max_dirty' \
  WHERE server='${SERVERNAME}' AND test='$TEST'"

# Cache server info in the test record to save CPU/memory upgrade history
SERVERCPU=`$RESULTPSQL -Atc "SELECT server_cpu FROM server WHERE server='${SERVERNAME}'"`
MEMGB=`$RESULTPSQL -Atc "SELECT server_mem_gb FROM server WHERE server='${SERVERNAME}'"`
if [ -z "${MEMGB}" ] ; then
  MEMGB=0
  echo WARNING:  missing entry "${SERVERNAME}" in server table
fi

# Save database size at the end to support init case
DBSIZE=`$TESTPSQL -A -t -c "select pg_database_size(current_database())"`
$RESULTPSQL -c "update tests set dbsize=${DBSIZE},server_mem_gb=${MEMGB},server_cpu='${SERVERCPU}' WHERE server='${SERVERNAME}' AND test='${TEST}'"

# Stats and buffer cache data are spoiled very fast by post processing, grab them as early as feasible
$TESTPSQL   -c "\copy (SELECT '$TEST', current_timestamp, stats_reset, numbackends, xact_commit, xact_rollback, blks_read, blks_hit, tup_returned, tup_fetched, tup_inserted, tup_updated, tup_deleted, conflicts, temp_files, temp_bytes, deadlocks, blk_read_time, blk_write_time, '$SERVERNAME' FROM pg_stat_database WHERE datname=current_database()) TO 'dbstat.csv' WITH DELIMITER ',' CSV HEADER"
$RESULTPSQL -c "\copy test_stat_database FROM 'dbstat.csv' WITH DELIMITER ',' CSV HEADER"
$TESTPSQL   -c "\copy (SELECT '$TEST', current_timestamp AS collected, schemaname AS nspname, relname AS tablename, indexrelname AS indexname, pg_relation_size(indexrelid) AS size, idx_blks_read AS rel_blks_read, idx_blks_hit AS rel_blks_hit,  '$SERVERNAME' FROM pg_statio_user_indexes UNION SELECT '$TEST', current_timestamp AS collected, schemaname AS nspname, relname AS tablename, NULL AS indexname, pg_relation_size(relid) AS size, heap_blks_read AS rel_blks_read, heap_blks_hit AS rel_blks_hit, '$SERVERNAME' FROM pg_statio_user_tables) TO 'statio.csv' WITH DELIMITER ',' CSV HEADER"
$RESULTPSQL -c "\copy test_statio FROM 'statio.csv' WITH DELIMITER ',' CSV HEADER"
$TESTPSQL   -c "\copy (SELECT '$SERVERNAME' AS server,'$TEST' AS test, N.nspname AS schemaname,c.relname,8192 * count(*) AS bytes, round(avg(usagecount)::numeric,2) AS avg_usage,max(usagecount) AS max_usage,isdirty FROM pg_class c LEFT JOIN pg_namespace N ON (N.oid = C.relnamespace) INNER JOIN pg_buffercache b ON b.relfilenode = c.relfilenode INNER JOIN pg_database d ON (b.reldatabase = d.oid AND d.datname = current_database()) WHERE NOT N.nspname IN ('pg_catalog','pg_toast','information_schema') GROUP BY N.nspname,c.relname,isdirty ) TO 'buffercache.csv' WITH DELIMITER ',' CSV HEADER"
$RESULTPSQL -c "\copy test_buffercache FROM 'buffercache.csv' WITH DELIMITER ',' CSV HEADER"

# Import test artifacts from results.  So far this only gives useful output
# for the pgbench init case and osm2pgsql.

$BASEDIR/util/pgbench-init-parse results.txt "${SCRIPT}" > results.json
$RESULTPSQL -f $BASEDIR/util/art-import.sql

# Export less time sensitive database statistics from pgbench into results database

$TESTPSQL -c "\copy (SELECT '${SERVERNAME}', '$TEST',
        name,
        setting,
        unit,
        source,
        boot_val,
        current_setting(name) AS value,
        CASE
            WHEN unit='B' THEN setting::numeric
            WHEN unit='8kB' THEN setting::numeric * 8192
            WHEN unit='kB' THEN setting::numeric * 1024
            WHEN unit='MB' THEN setting::numeric * 1024 * 1024
            WHEN unit='s' THEN setting::numeric
            WHEN unit='ms' THEN round(setting::numeric / 1000,3)
            WHEN unit='min' THEN setting::numeric * 60
            WHEN unit IS NULL AND vartype='integer' THEN setting::numeric
            WHEN unit IS NULL AND vartype='real' THEN setting::numeric
            END AS numeric_value,
        CASE
            WHEN unit IN ('B','8kB','kB','MB') THEN 'bytes'
            WHEN unit IN ('s','ms','min') THEN 'seconds'
            WHEN unit IN ('integer,real') THEN 'numeric'
            END AS numeric_unit
    FROM pg_settings
) TO 'pgset.csv' WITH DELIMITER ',' CSV HEADER"
$RESULTPSQL -c "\copy test_settings FROM 'pgset.csv' WITH DELIMITER ',' CSV HEADER"

$TESTPSQL   -c "\copy (SELECT '$SERVERNAME','$TEST',queryid,query,plans,total_plan_time,min_plan_time,max_plan_time,mean_plan_time,stddev_plan_time,calls,total_exec_time,min_exec_time,max_exec_time,mean_exec_time,stddev_exec_time,rows,shared_blks_hit,shared_blks_read,shared_blks_dirtied,shared_blks_written,local_blks_hit,local_blks_read,local_blks_dirtied,local_blks_written,temp_blks_read,temp_blks_written,blk_read_time,blk_write_time,wal_records,wal_fpi,wal_bytes FROM pg_stat_statements WHERE calls>1) TO 'statements.csv' WITH DELIMITER ',' CSV HEADER"
$RESULTPSQL -c "\copy test_statements FROM 'statements.csv' WITH DELIMITER ',' CSV HEADER"

# WAL values are initially saved as text because they're pulled from the test
# database--but the functions to turn them into numeric types is available
# only in the results one.
if [ "$SERVER_VERSION" -gt "100000" ]; then
END_WAL=`$TESTPSQL -A -t -F" " -c "select pg_current_wal_lsn()"`
else
END_WAL=`$TESTPSQL -A -t -F" " -c "select pg_current_xlog_location()"`
fi
$RESULTPSQL -q -c "UPDATE tests SET wal_written=wal_lsn('$END_WAL') - wal_lsn('$START_WAL') WHERE server='${SERVERNAME}' AND test=$TEST"

tail -n 40 results.txt

# Try to do the table cleanup before spoiling the database and OS cache by
# touching the results heavily.  That means before reading pgbench.log,
# which can be quite large on a fast server.
if [ "$SKIPINIT" -ne "1" ]; then
  echo Cleaning up database $TESTDB
  if [ "$PGBENCH_TABLES" -eq "1" ] ; then
    $TESTPSQL -c "truncate table ${TABLEPREFIX}history"
  fi
  VACUUMSTART=`$RESULTPSQL -A -t -c "SELECT now()"`
  $TESTPSQL -c "vacuum"
  $TESTPSQL -c "checkpoint"
  $RESULTPSQL -q -c "UPDATE tests SET cleanup=now() - '$VACUUMSTART'::timestamp WHERE server='${SERVERNAME}' AND test=$TEST"
fi

# Save pgbench log
# In the multi-threaded workers case, there may be a number of files, so copy
# them all in
cat pgbench_log.${P}* > pgbench.log
echo Worst latency results:

# TODO On Solaris, this may need to use /usr/xpg4/bin/tail instead
cat pgbench.log | cut -f 3 -d " " | sort -n | tail -n 5
tps=`egrep "(without initial connection time)|(including connections establishing)" results.txt | cut -d " " -f 3`
trans=`grep "number of transactions actually processed:" results.txt | cut -d":" -f 2 | cut -d "/" -f 1`

# Defaults for init and custom scripts that don't have transaction data.
if [ -z "$tps" ] ; then
  tps=0
fi
if [ -z "$trans" ] ; then
  trans=0
fi
$RESULTPSQL -q -c "UPDATE tests SET tps='$tps',trans='$trans', trans_latency='$trans' WHERE server='${SERVERNAME}' AND test=$TEST"

cp $BASEDIR/config config
cp $BASEDIR/params params
cp $BASEDIR/templates/test-index.html index.html
echo Importing and processing latency information

# Confirm we have an patched version of pgbench that has timestamps
TESTFORTS=`cat pgbench.log | head -n 1 | cut -d" " -f 6`
if [ -z "$TESTFORTS" ]; then
  if [ "$PGBENCH_TABLES" -eq "0" ] ; then
    echo Skipping pgbench latency analysis since those tables aren\'t present.
  elif [ "$PASSTHRU" -eq "1" ]; then
    echo Database passthrough requests like initialization don\'t provide pgbench latency.
  else
    echo "ERROR:  the pgbench used for this test is missing transaction"
    echo timestamps.  No latency information will be imported into
    echo the database and some plots will be missing.
    echo This may not be an error, some special features just don\'t create the right logs.
    fi
else
  # Import timestamp information
  # Format of csv file changes when --rates is used
  if [ -n "$2" ]; then
    $BASEDIR/log-to-csv_rates $TEST "${SERVERNAME}" < pgbench.log > timing.csv
    $RESULTPSQL -c "copy timing from stdin with csv" < timing.csv
    if [ "$TRIM" -gt 0 ] ; then
      $RESULTPSQL -c "UPDATE tests SET start_latency=start_time + (end_time - start_time)/10, end_latency=end_time - (end_time - start_time)/10 WHERE server='${SERVERNAME}' AND tests.test='$TEST';"
      $RESULTPSQL -c "WITH trimmed AS (SELECT start_time + (end_time - start_time)/10 as trim_start, end_time - (end_time - start_time)/10 as trim_end FROM tests WHERE server='${SERVERNAME}' AND test='$TEST') DELETE FROM timing WHERE server='${SERVERNAME}' AND test='$TEST' AND (ts<=(SELECT trim_start FROM trimmed) OR ts>=(SELECT trim_end FROM trimmed));"
      $RESULTPSQL -c "UPDATE tests SET trans_latency=(SELECT count(*) FROM timing WHERE server='${SERVERNAME}' AND tests.test='$TEST') WHERE server='${SERVERNAME}' AND tests.test='$TEST';"
    fi
    $RESULTPSQL -q -c "UPDATE tests set avg_latency=(select avg(latency) from timing where tests.test=timing.test), max_latency=(select max(latency)from timing where tests.test=timing.test), percentile_90_latency=(select latency from timing ti where tests.test=ti.test and latency IS NOT NULL ORDER BY latency OFFSET GREATEST(0,(SELECT trunc(0.90*count(*)) FROM timing WHERE latency IS NOT null) LIMIT 1)) WHERE server='${SERVERNAME}' AND tests.test='$TEST'"
  else
    $BASEDIR/log-to-csv $TEST "${SERVERNAME}" < pgbench.log > timing.csv
    $RESULTPSQL -c "copy timing from stdin with csv" < timing.csv
    if [ "$TRIM" -gt 0 ] ; then
      $RESULTPSQL -c "UPDATE tests SET start_latency=start_time + (end_time - start_time)/10, end_latency=end_time - (end_time - start_time)/10 WHERE server='${SERVERNAME}' AND tests.test='$TEST';"
      $RESULTPSQL -c "WITH trimmed AS (SELECT start_time + (end_time - start_time)/10 as trim_start, end_time - (end_time - start_time)/10 as trim_end FROM tests WHERE server='${SERVERNAME}' AND test='$TEST') DELETE FROM timing WHERE server='${SERVERNAME}' AND test='$TEST' AND (ts<=(SELECT trim_start FROM trimmed) OR ts>=(SELECT trim_end FROM trimmed));"
      $RESULTPSQL -c "UPDATE tests SET trans_latency=(SELECT count(*) FROM timing WHERE server='${SERVERNAME}' AND tests.test='$TEST') WHERE server='${SERVERNAME}' AND tests.test='$TEST';"
    fi
  fi
  $RESULTPSQL -q -c "UPDATE tests SET avg_latency=(select avg(latency) from timing where tests.test=timing.test), max_latency=(select max(latency)from timing where tests.test=timing.test), percentile_90_latency=(select latency from timing where tests.test=timing.test ORDER BY latency OFFSET GREATEST(0,(trunc(0.90*trans_latency))) LIMIT 1) WHERE server='${SERVERNAME}' AND tests.test='$TEST'"
fi

if [ "$OS" = "Linux" ] ; then
  $BASEDIR/dirty-plot < meminfo.log > dirtydata.txt
fi

# Save high level summary of latency broken into equal sized buckets of transactions
# 20 5% buckets
$RESULTPSQL -c "DROP TABLE buckets" >> /dev/null 2>&1
$RESULTPSQL -c "CREATE TABLE buckets AS SELECT trans,0.05 as bucket_size,0.0 as bucket_left,0.05*generate_series(1,20) AS bucket_right,0.0 as offset_left,0.0 as offset_right,0.0 as latency_left,0.0 as latency_right FROM tests WHERE server='${SERVERNAME}' AND test=$TEST"
$RESULTPSQL -c "UPDATE buckets SET bucket_left=bucket_right - bucket_size;"
$RESULTPSQL -c "UPDATE buckets SET offset_left=round(trans * bucket_left), offset_right=round(trans * bucket_right) - 1;"
$RESULTPSQL -c "UPDATE buckets SET latency_left=(SELECT latency FROM timing WHERE test=$TEST ORDER BY test,latency OFFSET GREATEST(0,buckets.offset_left) LIMIT 1), latency_right=(SELECT latency FROM timing WHERE test=$TEST ORDER BY test,latency OFFSET GREATEST(0,buckets.offset_right) LIMIT 1)"

$RESULTPSQL -c "ALTER TABLE buckets DROP COLUMN trans"
$RESULTPSQL -c "ALTER TABLE buckets DROP COLUMN bucket_size"
# DEBUG $RESULTPSQL -c "SELECT * FROM buckets ORDER BY bucket_left"
$RESULTPSQL -c "\copy (SELECT * FROM buckets) TO 'latency-20.csv' WITH DELIMITER ',' CSV HEADER"

# 100 1% buckets
if [ "$LARGEBUCKETS" -eq "1" ] ; then
$RESULTPSQL -c "DROP TABLE buckets"
$RESULTPSQL -c "CREATE TABLE buckets AS SELECT trans,0.01 as bucket_size,0.0 as bucket_left,0.01*generate_series(1,100) AS bucket_right,0.0 as offset_left,0.0 as offset_right,0.0 as latency_left,0.0 as latency_right FROM tests WHERE tests.test=$TEST"
$RESULTPSQL -c "UPDATE buckets SET bucket_left=bucket_right - bucket_size;"
$RESULTPSQL -c "UPDATE buckets SET offset_left=round(trans * bucket_left), offset_right=round(trans * bucket_right) - 1";
$RESULTPSQL -c "UPDATE buckets SET latency_left=(SELECT latency from timing WHERE test=$TEST ORDER BY test,latency OFFSET GREATEST(0,buckets.offset_left) LIMIT 1), latency_right=(SELECT latency FROM timing WHERE test=$TEST ORDER BY test,latency OFFSET GREATEST(0,buckets.offset_right) LIMIT 1)"
$RESULTPSQL -c "ALTER TABLE buckets DROP COLUMN trans"
$RESULTPSQL -c "ALTER TABLE buckets DROP COLUMN bucket_size"
$RESULTPSQL -c "\copy (SELECT * FROM buckets) TO 'latency-100.csv' WITH DELIMITER ',' CSV HEADER"
$RESULTPSQL -c "DROP TABLE buckets"
fi

# Save second resolution transaction summary and metrics summaries
$RESULTPSQL -c "\copy (SELECT extract(epoch FROM date_trunc('second',ts)) AS collected, count(*) AS samples, min(latency) AS min_latency, round(1000*avg(latency))/1000 AS avg_latency, max(latency) AS max_latency FROM timing WHERE server='${SERVERNAME}' AND test=$TEST GROUP BY date_trunc('second',ts) ORDER BY date_trunc('second',ts)) TO 'latency_1s.csv' WITH DELIMITER ',' CSV HEADER"

if [ -z "$BRIDGE_BEARER" ] ; then
    $BASEDIR/metrics2csv $TEST "${SERVERNAME}"
else
    $BASEDIR/util/cbmetrics-import $TEST "${SERVERNAME}"
fi

# Plot result graphs
if [ ! -z $GNUPLOT ] && [ -x $GNUPLOT ]; then
  $RESULTPSQL -A -t -F' ' -c "select extract(epoch from date_trunc('second',ts)),count(*) from timing where server='${SERVERNAME}' AND test=$TEST group by date_trunc('second',ts) order by date_trunc('second',ts)" > tpsdata.txt
  $GNUPLOT $BASEDIR/plots/tps.plot

  $RESULTPSQL -A -t -F' ' -c "select extract(epoch from ts),latency from timing WHERE server='${SERVERNAME}' AND test=$TEST" > latency.txt
  $GNUPLOT $BASEDIR/plots/latency.plot
  if [ -f dirtydata.txt ] ; then
    $GNUPLOT $BASEDIR/plots/dirty.plot
  fi
  # TODO If $GNUPLOT is set to something other than its default value from `which`, there
  # should be a command line option in csv2gnuplot that informs this program of that path.
  $BASEDIR/csv2gnuplot -i iostat.log -d iostat -t "pgbench iostat" --disks="$DISKLIST" -o iostat.png
  $BASEDIR/csv2gnuplot -i vmstat.log -d vmstat -t "pgbench vmstat" -o vmstat.png
else
  echo "gnuplot not found, skipping benchwarmer charts generation"
fi

# Now that we're done plotting and computing stats, wipe the low-level
# data we don't need anymore
$RESULTPSQL -q -c "truncate table timing"

# Save some configuration information about the server
CLIENTHOST=`hostname`
SERVERHOST="$TESTHOST"
if [ "$SERVERHOST" = "localhost" ]; then
  SERVERHOST="$CLIENTHOST"
fi
SETTINGS="pg_settings.txt"

# Write out system and PostgreSQL installation
echo Test results: > $SETTINGS
$RESULTPSQL -c "select script,clients,round(tps) as tps,1000*round(avg_latency)/1000 as avg_latency,1000*round(max_latency)/1000 as max_latency from tests where server='${SERVERNAME}' AND test=$TEST" | grep -v " row)" >> $SETTINGS
echo Server $SERVERHOST, client $CLIENTHOST >> $SETTINGS
echo >> $SETTINGS
echo "Server info and settings in postgresql.conf:" >> $SETTINGS
$TESTPSQL -At -c "select version();" | grep -v " row" >> $SETTINGS
$TESTPSQL -c "show data_directory" | grep -v " row" >> $SETTINGS
$TESTPSQL -c "select name,current_setting(name) from pg_settings where source='configuration file' and not name in ('DateStyle','lc_messages','lc_monetary','lc_numeric','lc_time','listen_addresses','log_directory','log_rotation_age','log_rotation_size','log_truncate_on_rotation');" | grep -v " rows)" >> $SETTINGS

# Operating system information
echo >> $SETTINGS
echo "benchmark client OS Configuration (may not be the same as the server)" >> $SETTINGS
uname -a >> $SETTINGS

# Linux OS release is likely to be in one or more of these files
for f in `ls /etc/lsb-release /etc/debian_version /etc/redhat-release 2>/dev/null` ; do
  echo $f: >> $SETTINGS
  cat $f >> $SETTINGS
  echo >> $SETTINGS
done

# Save version info on Mac OS X
if [ "$OS" = "Darwin" ] ; then
  sw_vers >> $SETTINGS
fi
echo >> $SETTINGS

if [ -d /proc/sys/vm/ ] ; then
  for f in `ls /proc/sys/vm/dirty_*` ; do
    S=`cat $f`
    echo  $f=$S >> $SETTINGS
  done
  echo >> $SETTINGS
fi

mount >> $SETTINGS

# Remove temporary files, don't complain if the files are missing e.g.
# because we skipped gnuplot'ing and generating these files
rm -f pgbench_log.${P}*
rm -f pgbench.log
rm -f timing.csv # DEBUG
rm -f tpsdata.txt
rm -f latency.txt
rm -f dirtydata.txt
rm -f dbstat.csv statio.csv
rm -f pgset.csv
rm -f statements.csv