X-Git-Url: http://git.onelab.eu/?p=nodemanager.git;a=blobdiff_plain;f=logger.py;h=767a04e6f8a3b930cfe70c868c1156ef925898b4;hp=dc65bc2f4403902199b891008b36dc504b19139a;hb=e57432c1dfdfeaa52cc32799e2abbc34b7704ce9;hpb=8a73c41fa38f330fd125d85d693d1390ea2477a4 diff --git a/logger.py b/logger.py index dc65bc2..767a04e 100644 --- a/logger.py +++ b/logger.py @@ -1,5 +1,3 @@ -# $Id$ -# $URL$ """A very simple logger that tries to be concurrency-safe.""" @@ -9,90 +7,74 @@ import traceback import subprocess import select -LOG_FILE = '/var/log/nm' -LOG_SLIVERS = '/var/log/getslivers.txt' +LOG_FILE = '/var/log/nodemanager' +LOG_SLIVERS = '/var/lib/nodemanager/getslivers.txt' +LOG_DATABASE = '/var/lib/nodemanager/database.txt' -# Thierry - trying to debug this for 4.2 # basically define 3 levels -LOG_NONE=0 -LOG_NODE=1 -LOG_VERBOSE=2 +LOG_NONE = 0 +LOG_NODE = 1 +LOG_VERBOSE = 2 # default is to log a reasonable amount of stuff for when running on operational nodes -LOG_LEVEL=1 +LOG_LEVEL = LOG_NODE def set_level(level): global LOG_LEVEL - assert level in [LOG_NONE,LOG_NODE,LOG_VERBOSE] - LOG_LEVEL=level + try: + assert level in [LOG_NONE, LOG_NODE, LOG_VERBOSE] + LOG_LEVEL = level + except: + logger.log("Failed to set LOG_LEVEL to %s" % level) def verbose(msg): - log('(v) '+msg,LOG_VERBOSE) + log('(v) ' + msg, LOG_VERBOSE) -def log(msg,level=LOG_NODE): +def log(msg, level=LOG_NODE): """Write to the log file if level >= current log level (default LOG_NODE).""" - if (level > LOG_LEVEL): + if level > LOG_LEVEL: return try: fd = os.open(LOG_FILE, os.O_WRONLY | os.O_CREAT | os.O_APPEND, 0600) - if not msg.endswith('\n'): msg += '\n' + if not msg.endswith('\n'): + msg += '\n' os.write(fd, '%s: %s' % (time.asctime(time.gmtime()), msg)) os.close(fd) except OSError: sys.stderr.write(msg) sys.stderr.flush() -#################### child processes -# avoid waiting until the process returns; -# that makes debugging of hanging children hard +date_width = 24 +def log_exc(msg="", name=None): + """Log traceback resulting from an exception.""" + printout = "" + if name: + printout += "%s: "%name + printout += "EXCEPTION caught <%s> \n" % msg + for frame in traceback.format_exc().split("\n"): + printout += (date_width+2)*" " + "%s\n" % frame + log(printout) -# time out in seconds - avoid hanging subprocesses - default is 5 minutes -default_timeout_minutes=5 +def log_trace(msg="", name=None): + """Log current stack""" + printout = "" + if name: + printout += "%s: " % name + printout += "LOGTRACE\n" + for frame in traceback.format_stack(): + printout += "..." + frame + log(printout) -def log_call(command,timeout=default_timeout_minutes*60,poll=0.3): - log('log_call: running command %s' % ' '.join(command)) - verbose('log_call: timeout %r s' % timeout) - verbose('log_call: poll %r s' % poll) - trigger=time.time()+timeout - try: - child = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, close_fds=True) - while True: - # see if anything can be read within the poll interval - (r,w,x)=select.select([child.stdout,child.stderr],[],[],poll) - # read and log it - for fd in r: - input=fd.read() - if input: log(input) - # is process over ? - returncode=child.poll() - # yes - if returncode != None: - # child is done and return 0 - if returncode == 0: - verbose('log_call: command completed %s' % ' '.join(command)) - break - # child has failed - else: - raise Exception("log_call: failed with returncode %d"%returncode) - # no : still within timeout ? - if time.time() >= trigger: - child.terminate() - raise Exception("log_call: terminated command - exceeded timeout %d s"%timeout) - except: log_exc('failed to run command %s' % ' '.join(command)) - -def log_exc(msg="",name=None): - """Log the traceback resulting from an exception.""" - if name: - log("%s: EXCEPTION caught <%s> \n %s" %(name, msg, traceback.format_exc())) - else: - log("EXCEPTION caught <%s> \n %s" %(msg, traceback.format_exc())) +########## snapshot data to a file # for some reason the various modules are still triggered even when the # data from PLC cannot be reached # we show this message instead of the exception stack instead in this case -def log_missing_data (msg,key): +def log_missing_data(msg,key): log("%s: could not find the %s key in data (PLC connection down?) - IGNORED"%(msg,key)) -def log_data_in_file (data, file, message=""): +def log_data_in_file(data, file, message="",level=LOG_NODE): + if level > LOG_LEVEL: + return import pprint, time try: f=open(file,'w') @@ -102,8 +84,74 @@ def log_data_in_file (data, file, message=""): pp=pprint.PrettyPrinter(stream=f,indent=2) pp.pprint(data) f.close() + verbose("logger:.log_data_in_file Owerwrote %s"%file) except: - log_verbose('log_data_in_file failed - file=%s - message=%r'%(file,message)) + log_exc('logger.log_data_in_file failed - file=%s - message=%r'%(file,message)) + +def log_slivers(data): + log_data_in_file(data, LOG_SLIVERS, "raw GetSlivers") +def log_database(db): + log_data_in_file(db, LOG_DATABASE, "raw database") -def log_slivers (data): - log_data_in_file (data, LOG_SLIVERS, "raw GetSlivers") +#################### child processes +# avoid waiting until the process returns; +# that makes debugging of hanging children hard + +class Buffer: + def __init__(self, message='log_call: '): + self.buffer = '' + self.message = message + + def add(self,c): + self.buffer += c + if c=='\n': + self.flush() + + def flush(self): + if self.buffer: + log(self.message + self.buffer) + self.buffer = '' + +# time out in seconds - avoid hanging subprocesses - default is 5 minutes +default_timeout_minutes = 5 + +# returns a bool that is True when everything goes fine and the retcod is 0 +def log_call(command, timeout=default_timeout_minutes*60, poll=1): + message=" ".join(command) + log("log_call: running command %s" % message) + verbose("log_call: timeout=%r s" % timeout) + verbose("log_call: poll=%r s" % poll) + trigger=time.time()+timeout + result = False + try: + child = subprocess.Popen(command, bufsize=1, + stdout=subprocess.PIPE, stderr=subprocess.STDOUT, close_fds=True) + buffer = Buffer() + while True: + # see if anything can be read within the poll interval + (r, w, x) = select.select([child.stdout], [], [], poll) + if r: + buffer.add(child.stdout.read(1)) + # is process over ? + returncode = child.poll() + # yes + if returncode != None: + buffer.flush() + # child is done and return 0 + if returncode == 0: + log("log_call:end command (%s) completed" % message) + result = True + break + # child has failed + else: + log("log_call:end command (%s) returned with code %d" %(message,returncode)) + break + # no : still within timeout ? + if time.time() >= trigger: + buffer.flush() + child.terminate() + log("log_call:end terminating command (%s) - exceeded timeout %d s"%(message,timeout)) + break + except: + log_exc("failed to run command %s" % message) + return result