gen_basho_bench_report.py

#!/usr/bin/python

import os
import sys
import re
from collections import defaultdict

variants = {
    "concurrent" : ["1000", "2000", "3000", "5000"],
    "key_generator" : ["uniform_int, 10000", "uniform_int, 10000000", "pareto_int, 10000", "pareto_int, 10000000"],
    "value_generator" : ["uniform_bin, 1000, 4512", "uniform_bin, 1000, 2512", "exponential_bin, 1000, 2000"],
    "operations": ["put, 1", "put, 5", "put, 10", "put, 10.*delete, 10", "put, 10.*update, 10"]
}


test_root = sys.argv[1]
test_dirs= sorted([os.path.join(test_root, o) for o in os.listdir(test_root) if os.path.isdir(os.path.join(test_root, o))])

known_errors = ["timeout", "disconnected", "other"]

print "<table border=1>"
print "<tr>"
print "  <th>image</th>"
for k in variants.keys():
    print "  <th>" + k + "</th>"

for e in known_errors:
    print "  <th>" + e + "</th>"
print "</tr>"
for d in test_dirs:
    config_path = os.path.join(d, "nsto2.config")
    sys.stderr.write("%s\n" % config_path)
    if not os.path.isfile(config_path):
        continue
    config = open(config_path, "r").read() 
    cur_vars = {}
    for (k, vs) in variants.iteritems():
        cur_vars.update([ (k,v) for v in vs if re.search(k + ".*" + v, config) ])

    errors_file = os.path.join(d, "errors.csv")
    
    errors = defaultdict(int)
    for line in open(errors_file, "rb"):
        if re.match("^\"error\",\"count\"$", line):
            continue
        m = re.match("^\".(.*),(.*).\",\"([0-9]*)\"$", line)
        #sys.stderr.write("%s\n" % line)
        error = m.group(2) if m.group(2) in known_errors else "other"
        errors[error] += int(m.group(3))

    print "<tr><td><a href=\"%s/summary.png\">%s/summary.png</a></td>" % (d,d)
    for k in variants.keys():
        print "  <td>" + cur_vars[k] + "</td>"
    for e in known_errors:
        print "  <td>" + str(errors[e]) + "</td>"
    print "</tr>"
print "</table>"