File locking and logging

[Kamus of Kadizhar]

Thanks to Robert Brewer, I got enough insight into logging to make it work....

Now I have another issue: file locking. Sorry if this is a very basic
question, but I can't find a handy reference anywhere that mentions this.

When a logger opens a log file for append, is it automatically locked so
other processes cannot write to it? And what happens if two or more
processes attempt to log an event at the same time?

Here's my situation. I have two or three workstations that will log an
event (the playing of a movie). The log file is NFS mounted and all
workstations will use the same log file. How is file locking implemented?
Or is it?

I've read through the various logger doc pages and this is never mentioned.

The logging code that works (for me at least) is this:

logging.basicConfig()
logFile = logging.handlers.RotatingFileHandler("/var/log/user/movies.log",'a',2000,4)
logFile.setLevel(logging.INFO)
formatter = logging.Formatter(hostname + ' %(asctime)s %(message)s',datefmt='%Y-%m-%d.%H:%M')
logFile.setFormatter(formatter)
logging.getLogger('').addHandler(logFile)
logging.warning(movieName)
logFile.flush()
logFile.close()

Any thoughts are appreciated....

Thanks,

--Kamus

 

[Vinay Sajip]

Thanks to Robert Brewer, I got enough insight into logging to make it work....

Now I have another issue: file locking. Sorry if this is a very basic
question, but I can't find a handy reference anywhere that mentions this.

When a logger opens a log file for append, is it automatically locked so
other processes cannot write to it? And what happens if two or more
processes attempt to log an event at the same time?

Here's my situation. I have two or three workstations that will log an
event (the playing of a movie). The log file is NFS mounted and all
workstations will use the same log file. How is file locking implemented?
Or is it?

No file locking is attempted by current logging handlers with respect
to other processes - an ordinary open() call is used. Within a given
Python process, concurrency support is is provided through threading
locks. If you need bullet-proof operation in the scenario where
multiple workstations are logging to the same file, you can do this
through having all workstations log via a SocketHandler to a
designated node, where you run a server process which locally logs to
file events received across the network. There is a working example of
this in the Python 2.4 docs.

Best regards,


Vinay Sajip