You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
rocksdb/tools/regression_test.sh

415 lines
14 KiB

Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
#!/bin/bash
# The RocksDB regression test script.
# REQUIREMENT: must be able to run make db_bench in the current directory
#
# This script will do the following things in order:
#
# 1. check out the specified rocksdb commit.
# 2. build db_bench using the specified commit
# 3. setup test directory $TEST_PATH. If not specified, then the test directory
# will be "/tmp/rocksdb/regression_test"
# 4. run set of benchmarks on the specified host
# (can be either locally or remotely)
# 5. generate report in the $RESULT_PATH. If RESULT_PATH is not specified,
# RESULT_PATH will be set to $TEST_PATH/current_time
#
# = Examples =
# * Run the regression test using rocksdb commit abcdef that outputs results
# and temp files in "/my/output/dir"
#r
# TEST_PATH=/my/output/dir COMMIT_ID=abcdef ./tools/regression_test.sh
#
# * Run the regression test on a remost host under "/my/output/dir" directory
# and stores the result locally in "/my/benchmark/results" using commit
# abcdef and with the rocksdb options specified in /my/path/to/OPTIONS-012345
# with 1000000000 keys in each benchmark in the regression test where each
# key and value are 100 and 900 bytes respectively:
#
# REMOTE_USER_AT_HOST=yhchiang@my.remote.host \
# TEST_PATH=/my/output/dir \
# RESULT_PATH=/my/benchmark/results \
# COMMIT_ID=abcdef \
# OPTIONS_FILE=/my/path/to/OPTIONS-012345 \
# NUM_KEYS=1000000000 \
# KEY_SIZE=100 \
# VALUE_SIZE=900 \
# ./tools/regression_test.sh
#
# = Regression test environmental parameters =
# DEBUG: If true, then the script will not checkout master and build db_bench
# if db_bench already exists
# Default: 0
# TEST_PATH: the root directory of the regression test.
# Default: "/tmp/rocksdb/regression_test"
# RESULT_PATH: the directory where the regression results will be generated.
# Default: "$TEST_PATH/current_time"
# REMOTE_USER_AT_HOST: If set, then test will run on the specified host under
# TEST_PATH directory and outputs test results locally in RESULT_PATH
# The REMOTE_USER_AT_HOST should follow the format user-id@host.name
# DB_PATH: the path where the rocksdb database will be created during the
# regression test. Default: $TEST_PATH/db
# WAL_PATH: the path where the rocksdb WAL will be outputed.
# Default: $TEST_PATH/wal
# OPTIONS_FILE: If specified, then the regression test will use the specified
# file to initialize the RocksDB options in its benchmarks. Note that
# this feature only work for commits after 88acd93 or rocksdb version
# later than 4.9.
# DELETE_TEST_PATH: If true, then the test directory will be deleted
# after the script ends.
# Default: 0
#
# = db_bench parameters =
# NUM_THREADS: The number of concurrent foreground threads that will issue
# database operations in the benchmark. Default: 16.
# NUM_KEYS: The key range that will be used in the entire regression test.
# Default: 1G.
# NUM_OPS: The number of operations (reads, writes, or deletes) that will
# be issued in EACH thread.
# Default: $NUM_KEYS / $NUM_THREADS
# KEY_SIZE: The size of each key in bytes in db_bench. Default: 100.
# VALUE_SIZE: The size of each value in bytes in db_bench. Default: 900.
# CACHE_SIZE: The size of RocksDB block cache used in db_bench. Default: 1G
# STATISTICS: If 1, then statistics is on in db_bench. Default: 0.
# COMPRESSION_RATIO: The compression ratio of the key generated in db_bench.
# Default: 0.5.
# HISTOGRAM: If 1, then the histogram feature on performance feature is on.
# STATS_PER_INTERVAL: If 1, then the statistics will be reported for every
# STATS_INTERVAL_SECONDS seconds. Default 1.
# STATS_INTERVAL_SECONDS: If STATS_PER_INTERVAL is set to 1, then statistics
# will be reported for every STATS_INTERVAL_SECONDS. Default 60.
# MAX_BACKGROUND_FLUSHES: The maxinum number of concurrent flushes in
# db_bench. Default: 4.
# MAX_BACKGROUND_COMPACTIONS: The maximum number of concurrent compactions
# in db_bench. Default: 16.
# SEEK_NEXTS: Controls how many Next() will be called after seek.
# Default: 10.
# SEED: random seed that controls the randomness of the benchmark.
# Default: $( date +%s )
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
#==============================================================================
# CONSTANT
#==============================================================================
TITLE_FORMAT="%40s,%25s,%30s,%7s,%9s,%8s,"
TITLE_FORMAT+="%10s,%13s,%14s,%11s,%12s,"
TITLE_FORMAT+="%7s,%11s,"
TITLE_FORMAT+="%9s,%10s,%10s,%10s,%10s,%10s,%5s"
TITLE_FORMAT+="\n"
DATA_FORMAT="%40s,%25s,%30s,%7s,%9s,%8s,"
DATA_FORMAT+="%10s,%13.0f,%14s,%11s,%12s,"
DATA_FORMAT+="%7s,%11s,"
DATA_FORMAT+="%9.0f,%10.0f,%10.0f,%10.0f,%10.0f,%10.0f,%5.0f"
DATA_FORMAT+="\n"
MAIN_PATTERN="$1""[[:blank:]]+:.*[[:blank:]]+([0-9\.]+)[[:blank:]]+ops/sec"
PERC_PATTERN="Percentiles: P50: ([0-9\.]+) P75: ([0-9\.]+) "
PERC_PATTERN+="P99: ([0-9\.]+) P99.9: ([0-9\.]+) P99.99: ([0-9\.]+)"
#==============================================================================
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
function main {
commit=${1:-"origin/master"}
test_root_dir=${TEST_PATH:-"/tmp/rocksdb/regression_test"}
init_arguments $test_root_dir
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
if [ $DEBUG -eq 0 ]; then
checkout_rocksdb $commit
build_db_bench
elif [ ! -f db_bench ]; then
build_db_bench
fi
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
setup_test_directory
# an additional dot indicates we share same env variables
run_db_bench "fillseqdeterministic" $NUM_KEYS 1 0
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
run_db_bench "readrandom"
run_db_bench "readwhilewriting"
run_db_bench "deleterandom" $((NUM_KEYS / 10 / $NUM_THREADS))
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
run_db_bench "seekrandom"
run_db_bench "seekrandomwhilewriting"
cleanup_test_directory $test_root_dir
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
echo ""
echo "Benchmark completed! Results are available in $RESULT_PATH"
}
############################################################################
function init_arguments {
K=1024
M=$((1024 * K))
G=$((1024 * M))
current_time=$(date +"%F-%H:%M:%S")
RESULT_PATH=${RESULT_PATH:-"$1/results/$current_time"}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
COMMIT_ID=`git log | head -n1 | cut -c 8-`
SUMMARY_FILE="$RESULT_PATH/SUMMARY.csv"
DB_PATH=${3:-"$1/db"}
WAL_PATH=${4:-""}
if [ -z "$REMOTE_USER_AT_HOST" ]; then
DB_BENCH_DIR=${5:-"."}
else
DB_BENCH_DIR=${5:-"$1/db_bench"}
fi
DEBUG=${DEBUG:-0}
SCP=${SCP:-"scp"}
SSH=${SSH:-"ssh"}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
NUM_THREADS=${NUM_THREADS:-16}
NUM_KEYS=${NUM_KEYS:-$((1 * G))} # key range
NUM_OPS=${NUM_OPS:-$(($NUM_KEYS / $NUM_THREADS))}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
KEY_SIZE=${KEY_SIZE:-100}
VALUE_SIZE=${VALUE_SIZE:-900}
CACHE_SIZE=${CACHE_SIZE:-$((1 * G))}
STATISTICS=${STATISTICS:-0}
COMPRESSION_RATIO=${COMPRESSION_RATIO:-0.5}
HISTOGRAM=${HISTOGRAM:-1}
NUM_MULTI_DB=${NUM_MULTI_DB:-1}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
STATS_PER_INTERVAL=${STATS_PER_INTERVAL:-1}
STATS_INTERVAL_SECONDS=${STATS_INTERVAL_SECONDS:-600}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
MAX_BACKGROUND_FLUSHES=${MAX_BACKGROUND_FLUSHES:-4}
MAX_BACKGROUND_COMPACTIONS=${MAX_BACKGROUND_COMPACTIONS:-16}
DELETE_TEST_PATH=${DELETE_TEST_PATH:-0}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
SEEK_NEXTS=${SEEK_NEXTS:-10}
SEED=${SEED:-$( date +%s )}
}
# $1 --- benchmark name
# $2 --- number of operations. Default: $NUM_KEYS
# $3 --- number of threads. Default $NUM_THREADS
# $4 --- use_existing_db. Default: 1
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
function run_db_bench {
# this will terminate all currently-running db_bench
find_db_bench_cmd="ps aux | grep db_bench | grep -v grep | grep -v aux | awk '{print \$2}'"
USE_EXISTING_DB=${4:-1}
ops=${2:-$NUM_OPS}
threads=${3:-$NUM_THREADS}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
echo ""
echo "======================================================================="
echo "Benchmark $1"
echo "======================================================================="
echo ""
db_bench_error=0
options_file_arg=$(setup_options_file)
echo "$options_file_arg"
db_bench_cmd="$DB_BENCH_DIR/db_bench \
--benchmarks=$1 --db=$DB_PATH --wal_dir=$WAL_PATH \
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
--use_existing_db=$USE_EXISTING_DB \
--disable_auto_compactions \
--threads=$threads \
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
--num=$NUM_KEYS \
--reads=$ops \
--writes=$ops \
--deletes=$ops \
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
--key_size=$KEY_SIZE \
--value_size=$VALUE_SIZE \
--cache_size=$CACHE_SIZE \
--statistics=$STATISTICS \
$options_file_arg \
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
--compression_ratio=$COMPRESSION_RATIO \
--histogram=$HISTOGRAM \
--seek_nexts=$SEEK_NEXTS \
--stats_per_interval=$STATS_PER_INTERVAL \
--stats_interval_seconds=$STATS_INTERVAL_SECONDS \
--max_background_flushes=$MAX_BACKGROUND_FLUSHES \
--num_multi_db=$NUM_MULTI_DB \
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
--max_background_compactions=$MAX_BACKGROUND_COMPACTIONS \
--seed=$SEED 2>&1"
kill_db_bench_cmd="pkill db_bench"
ps_cmd="ps aux"
if ! [ -z "$REMOTE_USER_AT_HOST" ]; then
echo "Running benchmark remotely on $REMOTE_USER_AT_HOST"
kill_db_bench_cmd="$SSH $REMOTE_USER_AT_HOST $kill_db_bench_cmd"
db_bench_cmd="$SSH $REMOTE_USER_AT_HOST $db_bench_cmd"
ps_cmd="$SSH $REMOTE_USER_AT_HOST $ps_cmd"
fi
## kill existing db_bench processes
eval "$kill_db_bench_cmd"
if [ $? -eq 0 ]; then
echo "Killed all currently running db_bench"
fi
## make sure no db_bench is running
# The following statement is necessary make sure "eval $ps_cmd" will success.
# Otherwise, if we simply check whether "$(eval $ps_cmd | grep db_bench)" is
# successful or not, then it will always be false since grep will return
# non-zero status when there's no matching output.
ps_output="$(eval $ps_cmd)"
exit_on_error $? "$ps_cmd"
# perform the actual command to check whether db_bench is running
grep_output="$(eval $ps_cmd | grep db_bench | grep -v grep)"
if [ "$grep_output" != "" ]; then
echo "Stopped regression_test.sh as there're still db_bench processes running:"
echo $grep_output
exit 1
fi
## run the db_bench
cmd="($db_bench_cmd || db_bench_error=1) | tee -a $RESULT_PATH/$1"
exit_on_error $?
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
echo $cmd
eval $cmd
exit_on_error $db_bench_error
update_report "$1" "$RESULT_PATH/$1" $ops $threads
}
function multiply {
echo "$1 * $2" | bc
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
}
# $1 --- name of the benchmark
# $2 --- the filename of the output log of db_bench
function update_report {
main_result=`cat $2 | grep $1`
exit_on_error $?
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
perc_statement=`cat $2 | grep Percentile`
exit_on_error $?
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
# Obtain micros / op
[[ $main_result =~ $MAIN_PATTERN ]]
ops_per_s=${BASH_REMATCH[1]}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
# Obtain percentile information
[[ $perc_statement =~ $PERC_PATTERN ]]
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
perc[0]=${BASH_REMATCH[1]} # p50
perc[1]=${BASH_REMATCH[2]} # p75
perc[2]=${BASH_REMATCH[3]} # p99
perc[3]=${BASH_REMATCH[4]} # p99.9
perc[4]=${BASH_REMATCH[5]} # p99.99
(printf "$DATA_FORMAT" \
$COMMIT_ID $1 $REMOTE_USER_AT_HOST $NUM_MULTI_DB $NUM_KEYS $KEY_SIZE $VALUE_SIZE \
$(multiply $COMPRESSION_RATIO 100) \
$3 $4 $CACHE_SIZE \
$MAX_BACKGROUND_FLUSHES $MAX_BACKGROUND_COMPACTIONS \
$ops_per_s \
$(multiply ${perc[0]} 1000) \
$(multiply ${perc[1]} 1000) \
$(multiply ${perc[2]} 1000) \
$(multiply ${perc[3]} 1000) \
$(multiply ${perc[4]} 1000) \
$DEBUG \
>> $SUMMARY_FILE)
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
exit_on_error $?
}
function exit_on_error {
if [ $1 -ne 0 ]; then
echo ""
echo "ERROR: Benchmark did not complete successfully."
if ! [ -z "$2" ]; then
echo "Failure command: $2"
fi
echo "Partial results are output to $RESULT_PATH"
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
echo "ERROR" >> $SUMMARY_FILE
exit $1
fi
}
function checkout_rocksdb {
echo "Checking out commit $1 ..."
git fetch --all
exit_on_error $?
git checkout $1
exit_on_error $?
}
function build_db_bench {
echo "Building db_bench ..."
make clean
exit_on_error $?
DEBUG_LEVEL=0 make db_bench -j32
exit_on_error $?
}
function run_remote {
if ! [ -z "$REMOTE_USER_AT_HOST" ]; then
cmd="$SSH $REMOTE_USER_AT_HOST $1"
else
cmd="$1"
fi
eval "$cmd"
exit_on_error $? "$cmd"
}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
function run_local {
eval "$1"
exit_on_error $?
}
function setup_options_file {
if ! [ -z "$OPTIONS_FILE" ]; then
if ! [ -z "$REMOTE_USER_AT_HOST" ]; then
options_file="$DB_BENCH_DIR/OPTIONS_FILE"
run_local "$SCP $OPTIONS_FILE $REMOTE_USER_AT_HOST:$options_file"
else
options_file="$OPTIONS_FILE"
fi
echo "--options_file=$options_file"
fi
echo ""
}
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
function setup_test_directory {
echo "Deleting old regression test directories and creating new ones"
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
run_remote "rm -rf $DB_PATH"
run_remote "rm -rf $DB_BENCH_DIR"
run_local "rm -rf $RESULT_PATH"
if ! [ -z "$WAL_PATH" ]; then
run_remote "rm -rf $WAL_PATH"
run_remote "mkdir -p $WAL_PATH"
fi
run_remote "mkdir -p $DB_PATH"
run_remote "mkdir -p $DB_BENCH_DIR"
run_remote "ls -l $DB_BENCH_DIR"
if ! [ -z "$REMOTE_USER_AT_HOST" ]; then
run_local "$SCP ./db_bench $REMOTE_USER_AT_HOST:$DB_BENCH_DIR/db_bench"
fi
run_local "mkdir -p $RESULT_PATH"
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
(printf $TITLE_FORMAT \
"commit id" "benchmark" "user@host" "num-dbs" "key-range" "key-size" \
"value-size" "compress-rate" "ops-per-thread" "num-threads" "cache-size" \
"flushes" "compactions" \
"ops-per-s" "p50" "p75" "p99" "p99.9" "p99.99" "debug" \
>> $SUMMARY_FILE)
exit_on_error $?
}
function cleanup_test_directory {
if [ $DELETE_TEST_PATH -ne 0 ]; then
echo "Clear old regression test directories and creating new ones"
run_remote "rm -rf $DB_PATH"
run_remote "rm -rf $WAL_PATH"
if ! [ -z "$REMOTE_USER_AT_HOST" ]; then
run_remote "rm -rf $DB_BENCH_DIR"
fi
run_remote "rm -rf $1"
else
echo "------------ DEBUG MODE ------------"
echo "DB PATH: $DB_PATH"
echo "WAL PATH: $WAL_PATH"
fi
Initial script for the new regression test Summary: This diff includes an initial script running a set of benchmarks for regression test. The script does the following things: checkout the specified rocksdb commit (or origin/master as default) make clean && DEBUG_LEVEL=0 make db_bench setup test directories run set of benchmarks and store results Currently, the script will run couple benchmarks, store all the benchmark output, extract micros per op and percentile information for each benchmark and store them in a single SUMMARY.csv file. The SUMMARY.csv will make the follow-up regression detection easier. In addition, the current script only takes env arguments to set important attributes of db_bench. Will follow-up with a patch that allows db_bench to construct options from an options file. Test Plan: NUM_KEYS=100 ./tools/regression_test.sh Sample SUMMARY.csv file: commit id, benchmark, ms-per-op, p50, p75, p99, p99.9, p99.99 7e23ddf575890510e7d2fc7a79b31a1bbf317917, fillseq, 15.28, 54.66, 77.14, 5000.00, 17900.00, 18483.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, overwrite, 13.54, 57.69, 86.39, 3000.00, 15600.00, 17013.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readrandom, 1.04, 0.80, 1.67, 293.33, 395.00, 504.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, readwhilewriting, 2.75, 1.01, 1.87, 200.00, 460.00, 485.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, deleterandom, 3.64, 48.12, 70.09, 200.00, 336.67, 347.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandom, 24.31, 391.87, 513.69, 872.73, 990.00, 1048.00 7e23ddf575890510e7d2fc7a79b31a1bbf317917, seekrandomwhilewriting, 14.02, 185.14, 294.15, 700.00, 1440.00, 1527.00 Reviewers: sdong, IslamAbdelRahman, kradhakrishnan, yiwu, andrewkr, gunnarku Reviewed By: gunnarku Subscribers: gunnarku, MarkCallaghan, andrewkr, dhruba, leveldb Differential Revision: https://reviews.facebook.net/D57597
9 years ago
}
############################################################################
main $@