rda-python-common 1.0.0__py3-none-any.whl

rda_python_common/PgSIG.py

@@ -0,0 +1,1157 @@
+#
+###############################################################################
+#
+#     Title : PgSIG.py
+#
+#    Author : Zaihua Ji,  zji@ucar.edu
+#      Date : 08/05/2020
+#   Purpose : python library module for start and control daemon process
+#
+# Work File : $DSSHOME/lib/python/PgSIG.py
+#    Github : https://github.com/NCAR/rda-shared-libraries.git
+# 
+###############################################################################
+#
+import os
+import re
+import sys
+import errno
+import signal
+import time
+from contextlib import contextmanager
+import PgLOG
+import PgDBI
+
+VUSERS = []  # allow users to start this daemon
+CPIDS = {}    # allow upto 'mproc' processes at one time for daemon
+CBIDS = {}    # allow upto 'bproc' background processes at one time for each child
+SDUMP = {
+  'DEF' : '/dev/null',
+  'ERR' : '',
+  'OUT' : ''
+}
+
+PGSIG = {
+   'QUIT'  : 0,    # 1 if QUIT signal received, quit server if no child
+   'MPROC' : 1,    # default number of multiple processes
+   'BPROC' : 1,    # default number of multiple background processes
+   'ETIME' : 20,   # default error waiting time (in seconds)
+   'WTIME' : 120,  # default waiting time (in seconds)
+   'DTIME' : 600,  # the daemon record refresh time (in seconds)
+   'RTIME' : 2400, # the web rda config unlocking and unconfigured system down waiting time (in seconds)
+   'CTIME' : 4800, # the lock cleaning & configued system down waiting time (in seconds)
+   'PPID'  : -1,   # 1 - server, (> 1) - child, 0 - non-daemon mode
+   'PID'   : 0,    # current process ID
+   'DNAME' : '',   # daemon name
+   'DSTR'  : '',   # string for daemon with user login name
+   'MTIME' : 0,    # maximum daemon running time in seconds, 0 for unlimited
+   'STIME' : 0,    # time the daemon is started
+   'STRTM' : '',   # string format of 'STIME'
+}
+
+#
+# add users for starting this daemon
+#
+def add_vusers(user = None, mores = None):
+
+   global VUSERS
+   if not user:
+      VUSERS = [] # clean all vusers
+   else:
+      VUSERS.append(user)
+   
+   if mores: VUSERS.extend(mores)
+   
+#
+# valid user for starting this daemon
+#
+def check_vuser(user, aname = None):
+
+   if user and VUSERS:
+      valid = 0;
+      for vuser in VUSERS:
+         if user == vuser:
+            valid = 1;
+            break
+
+      if valid == 0:
+         vuser = ', '.join(VUSERS)
+         PgLOG.pglog("{}: must be '{}' to run '{}'  in Daemon mode".format(user, vuser, aname), PgLOG.LGEREX) 
+
+#
+# turn this process into a daemon
+#
+# aname - application name, or daemon name
+# uname - user login name to started the application
+# mproc - upper limit of muiltiple child processes
+# wtime - waiting time (in seconds) for next process for the daemon
+# logon - turn on the logging if true
+# bproc - multiple background processes if > 1
+# mtime - maximum running time for the daemon if provided 
+#
+def start_daemon(aname, uname, mproc = 1, wtime = 120, logon = 0, bproc = 1, mtime = 0):
+
+   dstr = "Daemon '{}'{} on {}".format(aname, (" By {}".format(uname) if uname else ''), PgLOG.PGLOG['HOSTNAME'])
+
+   pid = check_daemon(aname, uname)
+   if pid:
+      PgLOG.pglog("***************** WARNNING **************************\n" +
+                  "**  {} is running as PID={}\n".format(dstr, pid) +
+                  "**  You need stop it before starting a new one!\n" +
+                  "*****************************************************" , PgLOG.WARNLG)
+      PgLOG.pglog("{} is already running as PID={}".format(dstr, pid), PgLOG.FRCLOG|PgLOG.MSGLOG)
+      sys.exit(0)
+
+   if mproc > 1: PGSIG['MPROC'] = mproc
+   if bproc > 1: PGSIG['BPROC'] = bproc
+   PGSIG['WTIME'] = get_wait_time(wtime, 120, "Polling Wait Time")
+   PGSIG['MTIME'] = get_wait_time(mtime, 0, "Maximum Running Time")
+
+   pid = process_fork(dstr)
+   cpid = pid if pid > 0 else os.getpid()
+   msg = "PID={},PL={},WI={}".format(cpid, PGSIG['MPROC'], PGSIG['WTIME'])
+   if PGSIG['MTIME']: msg += ",MT={}".format(PGSIG['MTIME'])
+   logmsg = "{}({}) started".format(dstr, msg)
+   if logon: logmsg += " With Logging On"
+   if pid > 0:
+      PgLOG.pglog(logmsg, PgLOG.WARNLG)
+      sys.exit(0)
+
+   os.setsid()
+   os.umask(0)
+
+   # setup to catch signals in daemon only
+   signal.signal(signal.SIGCHLD, clean_dead_child)
+   signal.signal(signal.SIGQUIT, signal_catch)
+   signal.signal(signal.SIGUSR1, signal_catch)
+   signal.signal(signal.SIGUSR2, signal_catch)
+   PGSIG['DSTR'] = dstr
+   PGSIG['DNAME'] = aname
+   PGSIG['STIME'] = int(time.time())
+   PGSIG['STRTM'] = PgLOG.current_datetime(PGSIG['STIME'])
+   PGSIG['PPID'] = 1
+   PGSIG['PID'] = cpid
+
+   sys.stdin = open(SDUMP['DEF'])
+   PgLOG.cmdlog("{} By {}".format(logmsg, PGSIG['STRTM']))
+
+   if logon:
+      PgLOG.PGLOG['LOGMASK'] &= ~(PgLOG.WARNLG|PgLOG.EMLLOG)   # turn off warn/email in daemon
+      set_dump()
+   else:
+      PgLOG.PGLOG['LOGMASK'] &= ~(PgLOG.LGWNEM)   # turn off log/warn/email in daemon
+      set_dump(SDUMP['DEF'])
+
+   PgLOG.PGLOG['BCKGRND'] = 1   # make sure the background flag is always on
+   PgDBI.pgdisconnect(1)   # disconnect database in daemon
+
+#
+# set dump output file
+#
+def set_dump(default = None):
+
+   errdump = PgLOG.get_environment("ERRDUMP", default)
+   outdump = PgLOG.get_environment("OUTDUMP", default)
+
+   if not errdump:
+      if not PgLOG.PGLOG['ERRFILE']:
+         PgLOG.PGLOG['ERRFILE'] = re.sub(r'\.log$', '.err', PgLOG.PGLOG['LOGFILE'], 1)
+      errdump = "{}/{}".format(PgLOG.PGLOG['LOGPATH'], PgLOG.PGLOG['ERRFILE'])
+
+   if errdump != SDUMP['ERR']:
+      sys.stderr = open(errdump, 'a')
+      SDUMP['ERR'] = errdump
+
+   if not outdump: outdump = "{}/{}".format(PgLOG.PGLOG['LOGPATH'], PgLOG.PGLOG['LOGFILE'])
+   if outdump != SDUMP['OUT']:
+      sys.stdout = open(outdump, 'a')
+      SDUMP['OUT'] = outdump
+
+#
+# stop daemon and log the ending info
+# 
+def stop_daemon(msg):
+
+   msg = " with " + msg if msg else ''
+   PgLOG.PGLOG['LOGMASK'] |= PgLOG.MSGLOG    # turn on logging before daemon stops
+   PgLOG.pglog("{} Started at {}, Stopped gracefully{} by {}".format(PGSIG['DSTR'], PGSIG['STRTM'], msg, PgLOG.current_datetime()), PgLOG.LOGWRN)
+
+#
+# check if a daemon is running already
+#
+# aname - application name for the daemon
+# uname - user login name who started the daemon
+# 
+# return the process id if yes and 0 if not
+#
+def check_daemon(aname, uname = None):
+
+   if uname:
+      check_vuser(uname, aname)
+      pcmd = "ps -u {} -f | grep {} | grep ' 1 '".format(uname, aname)
+      mp = "^\s*{}\s+(\d+)\s+1\s+".format(uname)
+   else:
+      pcmd = "ps -C {} -f | grep ' 1 '".format(aname)
+      mp = "^\s*\w+\s+(\d+)\s+1\s+"
+      
+   buf = PgLOG.pgsystem(pcmd, PgLOG.LOGWRN, 20+1024)
+   if buf:
+      cpid = os.getpid()
+      lines = buf.split('\n')
+      for line in lines:
+         ms = re.match(mp, line)
+         pid = int(ms.group(1)) if ms else 0
+         if pid > 0 and pid != cpid: return pid
+   
+   return 0
+
+#
+# check if an application is running already; other than the current processs
+#
+# aname - application name
+# uname - user login name who started the application
+# argv  - argument string
+#
+# return the process id if yes and 0 if not
+#
+def check_application(aname, uname = None, sargv = None):
+
+   if uname:
+      check_vuser(uname, aname)
+      pcmd = "ps -u {} -f | grep {} | grep -v ' grep '".format(uname, aname)
+      mp = "^\s*{}\s+(\d+)\s+(\d+)\s+.*{}\S*\s+(.*)$".format(uname, aname)
+   else:
+      pcmd = "ps -C {} -f".format(aname)
+      mp = "^\s*\w+\s+(\d+)\s+(\d+)\s+.*{}\S*\s+(.*)$".format(aname)
+
+   buf = PgLOG.pgsystem(pcmd, PgLOG.LOGWRN, 20+1024)
+   if not buf: return 0
+
+   cpids = [os.getpid(), os.getppid()]
+   pids = []
+   ppids = []
+   astrs = []
+   lines = buf.split('\n')
+   for line in lines:
+      ms = re.match(mp, line)
+      if not ms: continue
+      pid = int(ms.group(1))
+      ppid = int(ms.group(2))
+      if pid in cpids:
+         if ppid not in cpids: cpids.append(ppid)
+         continue
+      pids.append(pid)
+      ppids.append(ppid)
+      if sargv: astrs.append(ms.group(3))
+
+   pcnt = len(pids)
+   if not pcnt: return 0
+
+   i = 0
+   while i < pcnt:
+      pid = pids[i]
+      if pid and pid in cpids:
+         pids[i] = 0
+         ppid = ppids[i]
+         if ppid not in cpids: cpids.append(ppid)
+         i = 0
+      else:
+         i += 1
+
+   for i in range(pcnt):
+      pid = pids[i]
+      if pid and (not sargv or sargv.find(astrs[i]) > -1): return pid
+
+   return 0
+
+#
+# validate if the current process is a single one. Quit if not
+#
+def validate_single_process(aname, uname = None, sargv = None, logact = PgLOG.LOGWRN):
+
+      pid = check_application(aname, uname, sargv)
+      if pid:
+         msg = aname
+         if sargv: msg += ' ' + sargv
+         msg += ": already running as PID={} on {}".format(pid, PgLOG.PGLOG['HOSTNAME'])
+         if uname: msg += ' By ' + uname
+         PgLOG.pglog(msg + ', Quit Now', logact)
+         sys.exit(0)
+
+#
+# check how many processes are running for an application already
+#
+# aname - application name
+# uname - user login name who started the application
+# argv  - argument string
+#
+# return the the number of processes (exclude the child one)
+#
+def check_multiple_application(aname, uname = None, sargv = None):
+
+   if uname:
+      check_vuser(uname, aname)
+      pcmd = "ps -u {} -f | grep {} | grep -v ' grep '".format(uname, aname)
+      mp = "^\s*{}\s+(\d+)\s+(\d+)\s+.*{}\S*\s+(.*)$".format(uname, aname)
+   else:
+      pcmd = "ps -C {} -f".format(aname)
+      mp = "^\s*\w+\s+(\d+)\s+(\d+)\s+.*{}\S*\s+(.*)$".format(aname)
+
+   buf = PgLOG.pgsystem(pcmd, PgLOG.LOGWRN, 20+1024)
+   if not buf: return 0
+
+   dpids = [os.getpid(), os.getppid()]
+   pids = []
+   ppids = []
+   astrs = []
+   lines = buf.split('\n')
+   for line in lines:
+      ms = re.match(mp, line)
+      if not ms: continue
+      pid = int(ms.group(1))
+      ppid = int(ms.group(2))
+      if pid in dpids:
+         if ppid > 1 and ppid not in dpids: dpids.append(ppid)
+         continue
+      elif ppid in pids:
+         if pid not in dpids: dpids.append(pid)
+         continue
+      pids.append(pid)
+      ppids.append(ppid)
+      if sargv: astrs.append(ms.group(3))
+
+   pcnt = len(pids)
+   if not pcnt: return 0
+
+   i = 0
+   while i < pcnt:
+      pid = pids[i]
+      ppid = ppids[i]
+      if pid:
+         if pid in dpids:
+            if ppid > 1 and ppid not in dpids: dpids.append(ppid)
+            i = pids[i] = 0
+            continue
+         elif ppid in pids:
+            if pid not in dpids: dpids.append(pid)
+            i = pids[i] = 0
+            continue
+      i += 1
+
+   ccnt = 0
+   for i in range(pcnt):
+      if pids[i] and (not sargv or sargv.find(astrs[i]) > -1): ccnt += 1
+
+   return ccnt
+
+#
+# validate if the running processes reach the limit for the given app; Quit if yes
+#
+def validate_multiple_process(aname, plimit, uname = None, sargv = None, logact = PgLOG.LOGWRN):
+
+      pcnt = check_multiple_application(aname, uname, sargv)
+      if pcnt >= plimit:
+         msg = aname
+         if sargv: msg += ' ' + sargv
+         msg += ": already running in {} processes on {}".format(pcnt, PgLOG.PGLOG['HOSTNAME'])
+         if uname: msg += ' By ' + uname
+         PgLOG.pglog(msg + ', Quit Now', logact)
+         sys.exit(0)
+
+#
+# fork process
+#
+# return the defined result from call of fork
+#
+def process_fork(dstr):
+   
+   for i in range(10):   # try 10 times
+      try:
+         pid = os.fork()
+         return pid
+      except OSError as e:
+         if e.errno == errno.EAGAIN: 
+            os.sleep(5)
+         else:
+            PgLOG.pglog("{}: {}".format(dstr, str(e)), PgLOG.LGEREX)
+            break
+
+   PgLOG.pglog("{}: too many tries (10) for os.fork()".format(dstr), PgLOG.LGEREX)
+
+#
+# process the predefined signals
+#
+def signal_catch(signum, frame):
+
+   if PGSIG['PPID'] == 1:
+      tmp = 'Server'
+   elif PGSIG['PPID'] > 1:
+      tmp = 'Child'
+   else:
+      tmp = 'Process'
+
+   if signum == signal.SIGQUIT:
+      sname = "<{} - signal.SIGQUIT - Quit>".format(signum)
+   elif signum == signal.SIGUSR1:
+      linfo = 'Logging On'
+      if PgLOG.PGLOG['LOGMASK']&PgLOG.MSGLOG: linfo += ' & Debugging On'
+      sname = "<{} - signal.SIGUSR1 - {}>".format(signum, linfo)
+   elif signum == signal.SIGUSR2:
+      if PgLOG.PGLOG['DBGLEVEL']:
+         linfo = 'Logging off & Debugging Off'
+      else:
+         linfo = 'Logging Off'
+      sname = "<{} - signal.SIGUSR2 - {}>".format(signum, linfo)
+   else:
+      sname = "<{} - Signal Not Supports Yet>".format(signum)
+
+   dumpon = 1 if SDUMP['OUT'] and SDUMP['OUT'] != SDUMP['DEF'] else 0
+   if not dumpon: set_dump()
+   PgLOG.pglog("catches {} in {} {}".format(sname, tmp, PGSIG['DSTR']), PgLOG.LOGWRN|PgLOG.FRCLOG)
+
+   if signum == signal.SIGUSR1:
+      if PgLOG.PGLOG['LOGMASK']&PgLOG.MSGLOG:
+         PgLOG.PGLOG['DBGLEVEL'] = 1000  # turn logon twice
+      else:
+         PgLOG.PGLOG['LOGMASK'] |= PgLOG.MSGLOG   # turn on logging
+   elif signum == signal.SIGUSR2:
+      PgLOG.PGLOG['LOGMASK'] &= ~(PgLOG.MSGLOG)   # turn off logging
+      PgLOG.PGLOG['DBGLEVEL'] = 0   # turn off debugging
+      set_dump(SDUMP['DEF'])
+   else:
+     if not dumpon: set_dump(SDUMP['DEF'])
+     if signum == signal.SIGQUIT: PGSIG['QUIT'] = 1
+
+   if PGSIG['PPID'] <= 1 and len(CPIDS) > 0:  # passing signal to child processes
+      for pid in CPIDS: kill_process(pid, signum)
+
+#
+# wrapper function to call os.kill() logging caught error based on logact
+# return PgLOG.SUCCESS is success; PgLog.FAILURE if not
+#
+def kill_process(pid, signum, logact = 0):
+
+   try:
+      os.kill(pid, signum)
+   except Exception as e:
+      ret = PgLOG.FAILURE
+      if logact:
+         if type(signum) is int:
+            sigstr = str(signum)
+         else:
+            sigstr = "{}-{}".format(signum.name, int(signum))
+         PgLOG.pglog("Error pass signal {} to pid {}: {}".format(sigstr, pid, str(e)), logact)
+   else:
+      ret = PgLOG.SUCCESS
+
+   return ret
+ 
+#
+# wait child process to finish
+#
+def clean_dead_child(signum, frame):
+
+   live = 0
+
+   while True:
+      try:
+         dpid, status = os.waitpid(-1, os.WNOHANG)
+      except ChildProcessError as e:
+         break  # no child process any more
+      except Exception as e:
+         PgLOG.PGLOG("Error check child process: {}".format(str(e)), PgLOG.ERRLOG)
+         break
+      else:
+         if dpid == 0:
+           if live > 0: break   # wait twice if a process is still a live
+           live += 1
+         elif PGSIG['PPID'] < 2:
+            if dpid in CPIDS: del CPIDS[dpid]
+
+#
+# send signal to daemon and exit
+#
+def signal_daemon(sname, aname, uname):
+   
+   dstr = "Daemon '{}'{} on {}".format(aname, ((" By " + uname) if uname else ""), PgLOG.PGLOG['HOSTNAME'])
+   pid = check_daemon(aname, uname)
+
+   if pid > 0:
+      dstr += " (PID = {})".format(pid)
+      if re.match(r'^(quit|stop)$', sname, re.I):
+         signum = signal.SIGQUIT
+         msg = "QUIT"
+      elif re.match(r'^(logon|on)$', sname, re.I):
+         signum = signal.SIGUSR1
+         msg = "Logging ON"
+      elif re.match(r'^(logoff|off)$', sname, re.I):
+         signum = signal.SIGUSR2
+         msg = "Logging OFF"
+         PgLOG.PGLOG['DBGLEVEL'] = 0
+      else:
+         PgLOG.pglog("{}: invalid Signal for {}".format(sname, dstr), PgLOG.LGEREX)
+
+      if kill_process(pid, signum, PgLOG.LOGERR) == PgLOG.SUCCESS:
+         PgLOG.pglog("{}: signal sent to {}".format(msg, dstr), PgLOG.LOGWRN|PgLOG.FRCLOG)
+   else:
+      PgLOG.pglog(dstr + ": not running currently", PgLOG.LOGWRN|PgLOG.FRCLOG)
+
+   sys.exit(0)
+
+#
+# start a time child to run the command in case hanging
+#
+def timeout_command(cmd, logact = PgLOG.LOGWRN, cmdopt = 4):
+
+   if logact&PgLOG.EXITLG: logact &= ~PgLOG.EXITLG
+
+   PgLOG.pglog("> " + cmd, logact)
+   if start_timeout_child(cmd, logact):
+      PgLOG.pgsystem(cmd, logact, cmdopt)
+      sys.exit(0)
+
+#
+# start a timeout child process 
+#
+# return: 1 - in child, 0 - in parent
+#
+def start_timeout_child(msg, logact = PgLOG.LOGWRN):
+
+   pid = process_fork(msg)
+
+   if pid == 0:  # in child
+      signal.signal(signal.SIGQUIT, signal_catch)   # catch quit signal only
+      PGSIG['PPID'] = PGSIG['PID']
+      PGSIG['PID'] = pid = os.getpid()
+      PgLOG.cmdlog("Timeout child to " + msg, time.time(), 0)
+      PgDBI.pgdisconnect(0)   # disconnect database in child
+      return 1
+
+   # in parent
+   for i in range(PgLOG.PGLOG['TIMEOUT']):
+      if not check_process(pid): break
+      sys.sleep(2)
+
+   if check_process(pid):
+      msg += ": timeout({} secs) in CPID {}".format(2*PgLOG.PGLOG['TIMEOUT'], pid)
+      pids = kill_children(pid, 0)
+      sys.sleep(6)
+      if kill_process(pid, signal.SIGKILL, PgLOG.LOGERR): pids.insert(0, pid)
+
+      if pids: msg += "\nProcess({}) Killed".format(','.join(map(str, pids)))
+      PgLOG.pglog(msg, logact)
+
+   return 0
+
+#
+# kill children recursively start from the deepest and return the pids got killed
+#
+def kill_children(pid, logact = PgLOG.LOGWRN):
+
+   buf = PgLOG.pgsystem("ps --ppid {} -o pid".format(pid), logact, 20)
+   pids = []
+   if buf:
+      lines = buf.split('\n')
+      for line in lines:
+         ms = re.match(r'^\s*(\d+)', line)
+         if not ms: continue
+         cid = int(ms.group(1))
+         if not check_process(cid): continue
+         cids = kill_children(cid, logact)
+         if cids: pids = cids + pids
+         if kill_process(cid, signal.SIGKILL, logact) == PgLOG.SUCCESS: pids.insert(0, cid)
+
+   if logact and len(pids): PgLOG.pglog("Process({}) Killed".format(','.join(map(str, pids))), logact)
+
+   return pids
+
+#
+# start a child process
+# pname - unique process name 
+#
+def start_child(pname, logact = PgLOG.LOGWRN, dowait = 0):
+
+   global CBIDS
+   if PGSIG['MPROC'] < 2: return 1  # no need child process
+
+   if logact&PgLOG.EXITLG: logact &= ~PgLOG.EXITLG
+   if logact&PgLOG.MSGLOG: logact |= PgLOG.FRCLOG
+
+   if PGSIG['QUIT']:
+      return PgLOG.pglog("{} is in QUIT mode, cannot start CPID for {}".format(PGSIG['DSTR'], pname), logact)
+   elif len(CPIDS) >= PGSIG['MPROC']:
+      i = 0
+      while True:
+         pcnt = check_child(None, 0, logact)
+         if pcnt < PGSIG['MPROC']: break
+         if dowait:
+            show_wait_message(i, "{}-{}: wait any {} child processes".format(PGSIG['DSTR'], pname, pcnt), logact, dowait)
+            i += 1
+         else:
+            return PgLOG.pglog("{}-{}: {} child processes already running at {}".format(PGSIG['DSTR'], pname, pcnt, PgLOG.current_datetime()), logact)
+
+   if check_child(pname): return -1   # process is running already
+
+   pid = process_fork(PGSIG['DSTR'])
+   if pid:
+      CPIDS[pid] = pname   # record the child process id
+      PgLOG.pglog("{}: starts CPID {} for {}".format(PGSIG['DSTR'], pid, pname))
+   else:
+      signal.signal(signal.SIGQUIT, signal.SIG_DFL)   # turn off catch QUIT signal in child
+      PgLOG.PGLOG['LOGMASK'] &= ~PgLOG.WARNLG   # turn off warn in child
+      PGSIG['PPID'] = PGSIG['PID']
+      PGSIG['PID'] = pid = os.getpid()
+      PGSIG['MPROC'] = 1   # 1 in child process
+      CBIDS = {}         # empty backgroud proces info in case not
+      PGSIG['DSTR'] += ": CPID {} for {}".format(pid, pname)
+      PgLOG.cmdlog("CPID {} for {}".format(pid, pname))
+      PgDBI.pgdisconnect(0)  # disconnect database in child
+
+   return 1   # child started successfully
+
+#
+# get child process id for given pname 
+#
+def pname2cpid(pname):
+
+   for cpid in CPIDS:
+      if CPIDS[cpid] == pname: return cpid
+
+   return 0
+
+#
+# check one or all child processes if they are still running
+# pname - unique process name if given
+# pid - check this specified process id if given
+# dowait - 0 no wait, 1 wait all done, -1 wait only when all children are running
+# return the number of running processes if dowait == 0 or 1
+# return the number of none-running processes if dowait == -1
+#
+def check_child(pname, pid = 0, logact = PgLOG.LOGWRN, dowait = 0):
+
+   if PGSIG['MPROC'] < 2: return 0   # no child process
+
+   if logact&PgLOG.EXITLG: logact &= ~PgLOG.EXITLG
+   ccnt = i = 0
+   if dowait < 0: ccnt = 1 if (pid or pname) else PGSIG['MPROC']
+   while True:
+      pcnt = 0
+      if not pid and pname: pid = pname2cpid(pname)
+      if pid:
+         if check_process(pid):   # process is not done yet
+            if pname:
+               PgLOG.pglog("{}({}): Child still running".format(pname, pid), logact)
+            else:
+               PgLOG.pglog("{}: Child still running".format(pid), logact)
+            pcnt = 1
+         elif pid in CPIDS:
+            del CPIDS[pid]   # clean the saved info for the process
+      elif not pname:
+         cpids = list(CPIDS)
+         for cpid in cpids:
+            if check_process(cpid):  # process is not done yet
+               pcnt += 1
+            elif cpid in CPIDS:
+               del CPIDS[cpid]
+
+      if pcnt == 0 or dowait == 0 or pcnt < ccnt: break
+      show_wait_message(i, "{}: wait {}/{} child processes".format(PGSIG['DSTR'], pcnt, PGSIG['MPROC']), logact, dowait)
+      i += 1
+
+   return (ccnt - pcnt) if ccnt else pcnt
+
+#
+# start this process in none daemon mode
+#
+# aname - application name, or daemon name
+#  cact - short action name
+# uname - user login name to started the application
+# mproc - upper limit of muiltiple child processes
+# wtime - waiting time (in seconds) for next process
+#
+def start_none_daemon(aname, cact = None, uname = None, mproc = 1, wtime = 120, logon = 1, bproc = 1):
+
+   dstr = aname
+   if cact: dstr += " for Action " + cact
+   if uname:
+      dstr +=  " By " + uname
+      check_vuser(uname, aname)
+
+   signal.signal(signal.SIGQUIT, signal_catch)   # catch quit signal only
+   signal.signal(signal.SIGCHLD, clean_dead_child)
+   PGSIG['DSTR'] = dstr
+   PGSIG['DNAME'] = aname
+   PGSIG['PPID'] = 0
+   PGSIG['PID'] = os.getpid()
+   PGSIG['MPROC'] = mproc
+   PGSIG['BPROC'] = bproc
+   PgLOG.PGLOG['CMDTIME'] = PGSIG['WTIME'] = get_wait_time(wtime, 120, "Polling Wait Time")
+   if PGSIG['MPROC'] > 1:
+      PgLOG.cmdlog("starts non-daemon {}(ML={},WI={})".format(aname, PGSIG['MPROC'], PGSIG['WTIME']))
+      if not logon: PgLOG.PGLOG['LOGMASK'] &= ~PgLOG.MSGLOG  # turn off message logging
+
+#
+# check one process id other than the current one if it is still running
+# pid - specified process id
+# pmsg - process message if given
+#
+def check_process(pid):
+
+   buf = PgLOG.pgsystem("ps -p {} -o pid".format(pid), PgLOG.LGWNEX, 20)
+   if buf:
+      mp = r'^\s*{}$'.format(pid)
+      lines = buf.split('\n')
+      for line in lines:
+         if re.match(mp, line): return 1
+
+   return 0
+
+#
+# check a process id on give host
+#
+def check_host_pid(host, pid, pmsg = None, logact = PgLOG.LOGWRN):
+
+   cmd = 'rdaps'
+   if host: cmd += " -h " + host
+   cmd += " -p {}".format(pid)
+   buf = PgLOG.pgsystem(cmd, logact, 276)   # 4+16+256
+   if not buf: return (-1 if PgLOG.PGLOG['SYSERR'] else 0)
+   if pmsg: PgLOG.pglog(pmsg, logact&(~PgLOG.EXITLG))
+   return 1
+
+#
+# check one process id on a given host name if it is still running, with default timeout
+#   pid - specified process id
+#  ppid - specified parent process id
+# uname - user login name who started the daemon
+#  host - host name the pid supposed to be running on
+# aname - application name
+#  pmsg - process message if given
+#
+# return 1 if process is steal live, 0 died already, -1 error checking
+#
+def check_host_process(host, pid, ppid = 0, uname = None, aname = None, pmsg = None, logact = PgLOG.LOGWRN):
+
+   cmd = "rdaps"
+   if host: cmd += " -h " + host
+   if pid: cmd += " -p {}".format(pid)
+   if ppid: cmd += " -P {}".format(ppid)
+   if uname: cmd += " -u " + uname
+   if aname: cmd += " -a " + aname
+   buf = PgLOG.pgsystem(cmd, logact, 276)   # 4+16+256
+   if not buf: return (-1 if PgLOG.PGLOG['SYSERR'] else 0)
+   if pmsg: PgLOG.pglog(pmsg, logact&(~PgLOG.EXITLG))
+   return 1
+
+#
+# get a single slurm status record
+#
+def get_slurm_info(bcmd, logact = PgLOG.LOGWRN):
+
+   stat = {}
+   buf = PgLOG.pgsystem(bcmd, logact, 16)
+   if not buf: return stat
+
+   chkt = 1
+   lines = buf.split('\n')
+   for line in lines:
+      if chkt:
+         if re.match(r'^\s*JOBID\s', line, re.I):
+            ckeys = re.split(r'\s+', PgLOG.pgtrim(line))
+            kcnt = len(ckeys)
+            chkt = 0
+      else:
+         if re.match(r'^-----', line): continue
+         vals = re.split(r'\s+', PgLOG.pgtrim(line))
+         vcnt = len(vals)
+         if vcnt >= kcnt:
+            for i in range(kcnt):
+               ckeys[i] = ckeys[i].upper()
+               stat[ckeys[i]] = vals[i]
+         
+            if vcnt > kcnt:
+               for i in range(kcnt, vcnt):
+                  stat[ckeys[kcnt-1]] += ' ' + str(vals[i])
+         break
+
+   return stat
+
+#
+# get a single pbs status record via qstat
+#
+def get_pbs_info(qopts, multiple = 0, logact = 0, chkcnt = 1):
+
+   stat = {}
+   loop = 0
+   buf = None
+   while loop < chkcnt:
+      buf = PgLOG.pgsystem("qstat -n -w {}".format(qopts), logact, 16)
+      if buf: break
+      loop += 1
+      time.sleep(6)
+
+   if not buf: return stat
+
+   chkt = chkd = 1
+   lines = buf.split('\n')
+   for line in lines:
+      if chkt:
+         if re.match(r'^Job ID', line):
+            line = re.sub(r'^Job ID', 'JobID', line, 1)
+            ckeys = re.split(r'\s+', PgLOG.pgtrim(line))
+            ckeys[1] = 'UserName'
+            ckeys[3] = 'JobName'
+            ckeys[7] = 'Reqd' + ckeys[7]
+            ckeys[8] = 'Reqd' + ckeys[7]
+            ckeys[9] = 'State'
+            ckeys[10] = 'Elap' + ckeys[7]
+            ckeys.append('Node')
+            kcnt = len(ckeys)
+            if multiple:
+               for i in range(kcnt):
+                  stat[ckeys[i]] = []
+            chkt = 0
+      elif chkd:
+         if re.match(r'^-----', line): chkd = 0
+      else:
+         vals = re.split(r'\s+', PgLOG.pgtrim(line))
+         if len(vals) == kcnt:
+            ms = re.match(r'^(\d+)', vals[0])
+            if ms: vals[0] = ms.group(1)
+            if multiple:
+               for i in range(kcnt):
+                  stat[ckeys[i]].append(vals[i])
+            else:
+               for i in range(kcnt):
+                  stat[ckeys[i]] = vals[i]
+               break
+
+   return stat
+
+#
+# get multiple slurn status record
+#
+def get_slurm_multiple(bcmd, logact = PgLOG.LOGWRN):
+
+   buf = PgLOG.pgsystem(bcmd, logact, 16)
+   if not buf: return 0
+
+   stat = {}
+   j = 0
+   chkt = chkd = 1
+   lines = buf.split('\n')
+   for line in lines:
+      if chkt:
+         if re.match(r'^\s*JOBID\s', line, re.I):
+            ckeys = re.split(r'\s+', PgLOG.pgtrim(line))
+            kcnt = len(ckeys)
+            for i in range(kcnt):
+               ckeys[i] = ckeys[i].upper()
+               stat[ckeys[i]] = []
+            chkt = 0
+      elif chkd:
+         if re.match(r'^-----', line): chkd = 0
+      else:
+         vals = re.split(r'\s+', PgLOG.pgtrim(line))
+         vcnt = len(vals)
+         if vcnt >= kcnt:
+            for i in range(kcnt):
+               stat[ckeys[i]].append(vals[i])
+         
+            if vcnt > kcnt:
+               for i in range(kcnt, vcnt):
+                  stat[ckeys[kcnt-1]][j] += ' ' + str(vals[i])
+            j += 1
+
+   return stat if j else 0
+
+#
+# check status of a slurm batch id
+#   bid - specified batch id
+#
+# return hash of batch status, 0 if cannot check any more
+#
+def check_slurm_status(bid, logact = PgLOG.LOGWRN):
+
+   return get_slurm_info("sacct -o jobid,user,totalcpu,elapsed,ncpus,state,jobname,nodelist -j {}".format(bid), logact)
+
+#
+# check status of a pbs batch id
+#   bid - specified batch id
+#
+# return hash of batch status, 0 if cannot check any more
+#
+def check_pbs_status(bid, logact = PgLOG.LOGWRN):
+
+   stat = {}
+   buf = PgLOG.pgsystem("qhist -w -j {}".format(bid), logact, 20)
+   if not buf: return stat
+
+   chkt = 1
+   lines = buf.split('\n')
+   for line in lines:
+      if chkt:
+         if re.match(r'^Job', line):
+            line = re.sub(r'^Job ID', 'JobID', line, 1)
+            line = re.sub(r'Finish Time', 'FinishTime', line, 1)
+            line = re.sub(r'Req Mem', 'ReqMem', line, 1)
+            line = re.sub(r'Used Mem\(GB\)', 'UsedMem(GB)', line, 1)
+            line = re.sub(r'Avg CPU \(%\)', 'AvgCPU(%)', line, 1)
+            line = re.sub(r'Elapsed \(h\)', 'WallTime(h)', line, 1)
+            line = re.sub(r'Job Name', 'JobName', line, 1)
+            ckeys = re.split(r'\s+', PgLOG.pgtrim(line))
+            ckeys[1] = 'UserName'
+            kcnt = len(ckeys)
+            chkt = 0
+      else:
+         vals = re.split(r'\s+', PgLOG.pgtrim(line))
+         for i in range(kcnt):
+            stat[ckeys[i]] = vals[i]
+         break
+   
+   return stat
+
+#
+# check if a slurm batch id is live
+#   bid - specified batch id
+#
+# return 1 if process is steal live, 0 died already or error checking
+#
+def check_slurm_process(bid, pmsg = None, logact = PgLOG.LOGWRN):
+
+   stat = get_slurm_info("squeue -l -j {}".format(bid), logact)
+
+   if stat:
+      ms = re.match(r'^(RUNNING|PENDING|SUSPENDE|COMPLETI|CONFIGUR|REQUEUE_)$', stat['STATE'])
+      if ms:
+         if pmsg: PgLOG.pglog("{}, STATE={}".format(pmsg, ms.group(1)), logact&~PgLOG.EXITLG)
+         return 1
+      else:
+         return 0
+
+   return -1
+
+#
+# check if a pbs batch id is live
+#   bid - specified batch id
+#
+# return 1 if process is steal live, 0 died already or error checking
+#
+def check_pbs_process(bid, pmsg = None, logact = PgLOG.LOGWRN):
+
+   stat = get_pbs_info(bid, 0, logact)
+
+   ret = -1
+   if stat:
+      ms = re.match(r'^(B|R|Q|S|H|W|X)$', stat['State'])
+      if ms:
+         if pmsg: pmsg += ", STATE='{}' and returns 1".format(ms.group(1))
+         ret = 1
+      else:
+         if pmsg: pmsg += ", STATE='{}' and returns 0".format(stat['State'])
+         ret = 0
+   elif pmsg:
+      pmsg += ", Process Not Exists and returns -1"
+      
+   if pmsg: PgLOG.pglog(pmsg, logact&~PgLOG.EXITLG)
+
+   return ret
+
+#
+# get wait time
+#
+def get_wait_time(wtime, default, tmsg):
+   
+   if not wtime: wtime = default   # use default time
+
+   if type(wtime) is int: return wtime
+   if re.match(r'^(\d*)$', wtime): return int(wtime)
+
+   ms = re.match(r'^(\d*)([DHMS])$', wtime, re.I)
+   if ms:
+      ret = int(ms.group(1))
+      unit = ms.group(2)
+   else:
+      PgLOG.pglog("{}: '{}' NOT in (D,H,M,S)".format(wtime, tmsg), PgLOG.LGEREX)
+
+   if unit != 'S':
+      ret *= 60   # seconds in a minute
+      if unit != 'M':
+         ret *= 60   # minutes in an hour
+         if unit != 'H':
+            ret *= 24   # hours in a day
+
+   return ret   # in seconds
+
+#
+# start a background process and record its id; check PgLOG.pgsystem() in PgLOG.pm for
+# valid cmdopt values
+#
+def start_background(cmd, logact = PgLOG.LOGWRN, cmdopt = 5, dowait = 0):
+
+   if PGSIG['BPROC'] < 2: return PgLOG.pgsystem(cmd, logact, cmdopt)  # no background
+
+   act = logact&(~PgLOG.EXITLG)
+   if act&PgLOG.MSGLOG: act |= PgLOG.FRCLOG  # make sure background calls always logged
+
+   if len(CBIDS) >= PGSIG['BPROC']:
+      i = 0
+      while True:
+         bcnt = check_background(None, 0, act)
+         if bcnt < PGSIG['BPROC']: break
+         if dowait:
+            show_wait_message(i, "{}-{}: wait any {} background calls".format(PGSIG['DSTR'], cmd, bcnt), act, dowait)
+            i += 1
+         else:
+            return PgLOG.pglog("{}-{}: {} background calls already at {}".format(PGSIG['DSTR'], cmd, bcnt, PgLOG.current_datetime()), act)
+
+   cmdlog = (act if cmdopt&1 else PgLOG.WARNLG)
+   if cmdopt&8:
+      PgLOG.cmdlog("starts '{}'".format(cmd), None, cmdlog)
+   else:
+      PgLOG.pglog("{}({})-{} >{} &".format(PgLOG.PGLOG['HOSTNAME'], os.getpid(), PgLOG.current_datetime(), cmd), cmdlog)
+   bckcmd = cmd
+   if cmdopt&2:
+      bckcmd += " >> {}/{}".format(PgLOG.PGLOG['LOGPATH'], PgLOG.PGLOG['LOGFILE'])
+
+   if cmdopt&4:
+      if not PgLOG.PGLOG['ERRFILE']:
+         PgLOG.PGLOG['ERRFILE'] = re.sub(r'\.log$', '.err', PgLOG.PGLOG['LOGFILE'], 1)
+      bckcmd += " 2>> {}/{}".format(PgLOG.PGLOG['LOGPATH'], PgLOG.PGLOG['ERRFILE'])
+
+   bckcmd += " &"
+   os.system(bckcmd)
+   return record_background(cmd, logact)
+
+#
+# get background process id for given bcmd 
+#
+def bcmd2cbid(bcmd):
+
+   for cbid in CBIDS:
+      if CBIDS[cbid] == bcmd: return cbid
+
+   return 0
+
+#
+# check one or all child processes if they are still running
+# bid - check this specified background process id if given
+# return the number of processes are still running
+#
+def check_background(bcmd, bid = 0, logact = PgLOG.LOGWRN, dowait = 0):
+
+   if PGSIG['BPROC'] < 2: return 0  # no background process
+
+   if logact&PgLOG.EXITLG: logact &= ~PgLOG.EXITLG
+   if not bid and bcmd: bid = bcmd2cbid(bcmd)
+   bcnt = i = 0
+   while True:
+      if bid:
+         if check_process(bid):  # process is not done yet
+            if bcmd:
+               PgLOG.pglog("{}({}): Background process still running".format(bcmd, bid), logact)
+            else:
+               PgLOG.pglog("{}: Background process still running".format(bid), logact)
+            bcnt = 1
+         elif bid in CBIDS:
+            del CBIDS[bid]   # clean the saved info for the process
+      elif not bcmd:
+         for bid in CBIDS:
+            if check_process(bid):  # process is not done yet
+               bcnt += 1
+            else:
+               del CBIDS[bid]
+
+      if not (bcnt and dowait): break
+      show_wait_message(i, "{}: wait {}/{} background processes".format(PGSIG['DSTR'], bcnt, PGSIG['MPROC']), logact, dowait)
+      i += 1
+      bcnt = 0
+
+   return bcnt
+
+#
+# check and record process id for background command; return 1 if success full;
+# 0 otherwise; -1 if done already
+#
+def record_background(bcmd, logact = PgLOG.LOGWRN):
+
+   ms = re.match(r'^(\S+)', bcmd)
+   if ms:
+      aname = ms.group(1)
+   else:
+      aname = bcmd
+
+   mp = r"^\s*(\S+)\s+(\d+)\s+1\s+.*{}(.*)$".format(aname)
+   pc = "ps -u {},{} -f | grep ' 1 ' | grep {}".format(PgLOG.PGLOG['CURUID'], PgLOG.PGLOG['RDAUSER'], aname)
+   for i in range(2):
+      buf = PgLOG.pgsystem(pc, logact, 20+1024)
+      if buf:
+         lines = buf.split('\n')
+         for line in lines:
+            ms = re.match(mp, line)
+            if not ms: continue
+            (uid, sbid, acmd) = ms.groups()
+            bid = int(sbid)
+            if bid in CBIDS: return -1
+            if uid == PgLOG.PGLOG['RDAUSER']:
+               acmd = re.sub(r'^\.(pl|py)\s+', '', acmd, 1)
+            if re.match(r'^{}{}'.format(aname, acmd), bcmd): continue
+            CBIDS[bid] = bcmd
+            return 1
+      time.sleep(2)
+
+   return 0
+
+#
+# sleep for given period for the daemon, stops if maximum running time reached
+#
+def sleep_daemon(wtime = 0, mtime = None):
+
+   if not wtime: wtime = PGSIG['WTIME']
+   if mtime is None: mtime = PGSIG['MTIME']
+   
+   if mtime > 0:
+      rtime = int(time.time()) - PGSIG['STIME']
+      if rtime >= mtime:
+         PGSIG['QUIT'] = 1
+         wtime = 0
+
+   if wtime: time.sleep(wtime)
+   return wtime
+
+#
+# show wait message every dintv and then sleep for PGSIG['WTIME']
+#
+def show_wait_message(loop, msg, logact = PgLOG.LOGWRN, dowait = 0):
+
+   if loop > 0 and (loop%30) == 0:
+      PgLOG.pglog("{} at {}".format(msg, PgLOG.current_datetime()), logact)
+
+   if dowait: time.sleep(PGSIG['WTIME'])
+
+#
+# register a time out function to raise a time out error
+#
+@contextmanager
+def pgtimeout(seconds = 0, logact = 0):
+   
+   if not seconds: seconds = PgLOG.PGLOG['TIMEOUT']
+   signal.signal(signal.SIGALRM, raise_pgtimeout)
+   signal.alarm(seconds)
+   try:
+      yield
+   except TimeoutError as e:
+      pass
+   finally:
+      signal.signal(signal.SIGALRM, signal.SIG_IGN)
+
+def raise_pgtimeout(signum, frame):
+    raise TimeoutError
+
+def timeout_func():
+    # Add a timeout block.
+    with pgtimeout(1):
+        print('entering block')
+        import time
+        time.sleep(10)
+        print('This should never get printed because the line before timed out')