matcher_top_jstack.py

#Script has one input paramerter, folder where are the top and jstack files
#Script will get all the content from PATH, look for files with topdashHOutput name, check the CPU_USAGE_COLUMN and when the value is higher than HIGH_CPU will chck in the jstak files for the Hex code
import rlcompleter, readline
import sys
import subprocess

print 'Searching in ... :'+ sys.argv[1]

readline.parse_and_bind('tab:complete')
PATH = sys.argv[1]
HIGH_CPU = 90
CPU_USAGE_COLUMN = 8
PID_COLUMN = 0
STRINGS_TO_IGNORE = ['GC task thread',]

# get folder info
task = subprocess.Popen('ls '+PATH,
                        shell=True,
                        stdout=subprocess.PIPE)
directory_raw_content = task.stdout.read()
directory_content = directory_raw_content.split('\n')


pids_to_check = []
######Obtain high cpu from top..   topdashHOutput.XXXX.txt
print directory_content
for topdashfile in directory_content:
  #if 'mytest' in topdashfile:
  if 'topdashHOutput' in topdashfile:
    print "checking high cpu in "+topdashfile
    f_topdashfile = open(PATH+topdashfile)
    line = f_topdashfile.readline()
    while line:
      fields = line.split()
      if len(fields) >= CPU_USAGE_COLUMN:
        #check if it's a float
        try:
            #if the number it's less 100 to avoid bad headers from top
            if float(fields[CPU_USAGE_COLUMN]) > HIGH_CPU and float(fields[CPU_USAGE_COLUMN]) < 101:
              pids_to_check.append(fields[PID_COLUMN])
        except ValueError:
            None
        #print fields[CPU_USAGE_COLUMN]
      line = f_topdashfile.readline()
    f_topdashfile.close()

hex_pid_to_check = []
#hexadecimalize it...
for pid in pids_to_check:
   hex_pid_to_check.append(hex(int(pid)))

coincidences = []
######Obtain jstack
for entry in directory_content:
  if 'jstack' in entry:
    print 'Analizing...'+entry
    jstack_file = open(PATH+entry, "r")
    jstack_line = jstack_file.readline()
    while jstack_line:
      if 'nid' in jstack_line:
        #black magiC that check all the array of HEX pids in each line of the jstack
         if [e for e in hex_pid_to_check if e in jstack_line]:
           #black magiC that exclude known strings
           if not [l for l in STRINGS_TO_IGNORE if l in jstack_line]:
             #I would like to retrieve here the whole stack
             fullstack = jstack_line
             jstack_line = jstack_file.readline()
             #this is ugle as Hell
             while jstack_line and 'Locked ownable synchronizers:' not in jstack_line:
               fullstack += jstack_line
               #print "we are seeking: ######3 :"+jstack_line+" 3###"
               jstack_line = jstack_file.readline()
             coincidences.append(fullstack)
      jstack_line = jstack_file.readline()
    jstack_file.close()

for i in coincidences:
  print i