X-Git-Url: http://git.onelab.eu/?a=blobdiff_plain;f=logger.py;h=dc65bc2f4403902199b891008b36dc504b19139a;hb=8a73c41fa38f330fd125d85d693d1390ea2477a4;hp=3411df50a8e4bd0b35521ca3c4c1a4890875601a;hpb=aac3e5d7c3443d6e1cb33525aefad35be5fe077a;p=nodemanager.git

diff --git a/logger.py b/logger.py
index 3411df5..dc65bc2 100644
--- a/logger.py
+++ b/logger.py
@@ -1,27 +1,109 @@
-import fcntl
-import os
-import subprocess
+# $Id$
+# $URL$
+
+"""A very simple logger that tries to be concurrency-safe."""
+
+import os, sys
 import time
 import traceback
+import subprocess
+import select
+
+LOG_FILE = '/var/log/nm'
+LOG_SLIVERS = '/var/log/getslivers.txt'
+
+# Thierry - trying to debug this for 4.2
+# basically define 3 levels
+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
 
-from config import LOG_FILE
+def set_level(level):
+    global LOG_LEVEL
+    assert level in [LOG_NONE,LOG_NODE,LOG_VERBOSE]
+    LOG_LEVEL=level
 
+def verbose(msg):
+    log('(v) '+msg,LOG_VERBOSE)
 
-def log(msg):
-    """Write <msg> to the log file."""
-    # the next three lines ought to be an atomic operation but aren't
-    fd = os.open(LOG_FILE, os.O_WRONLY | os.O_CREAT | os.O_APPEND, 0600)
-    flags = fcntl.fcntl(fd, fcntl.F_GETFD)
-    fcntl.fcntl(fd, fcntl.F_SETFD, flags | fcntl.FD_CLOEXEC)
-    if not msg.endswith('\n'): msg += '\n'
-    os.write(fd, '%s: %s' % (time.asctime(time.gmtime()), msg))
-    os.close(fd)
+def log(msg,level=LOG_NODE):
+    """Write <msg> to the log file if level >= current log level (default LOG_NODE)."""
+    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'
+        os.write(fd, '%s: %s' % (time.asctime(time.gmtime()), msg))
+        os.close(fd)
+    except OSError:
+        sys.stderr.write(msg)
+        sys.stderr.flush()
 
-def log_call(*args):
-    log('running command %s' % ' '.join(args))
-    try: subprocess.call(args)
-    except: log_exc()
+#################### child processes
+# avoid waiting until the process returns; 
+# that makes debugging of hanging children hard
 
-def log_exc():
+# time out in seconds - avoid hanging subprocesses - default is 5 minutes
+default_timeout_minutes=5
+
+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."""
-    log(traceback.format_exc())
+    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()))
+
+# 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):
+    log("%s: could not find the %s key in data (PLC connection down?) - IGNORED"%(msg,key))
+
+def log_data_in_file (data, file, message=""):
+    import pprint, time
+    try:
+        f=open(file,'w')
+        now=time.strftime("Last update: %Y.%m.%d at %H:%M:%S %Z", time.localtime())
+        f.write(now+'\n')
+        if message: f.write('Message:'+message+'\n')
+        pp=pprint.PrettyPrinter(stream=f,indent=2)
+        pp.pprint(data)
+        f.close()
+    except:
+        log_verbose('log_data_in_file failed - file=%s - message=%r'%(file,message))
+
+def log_slivers (data):
+    log_data_in_file (data, LOG_SLIVERS, "raw GetSlivers")