Python, want logging with log rotation and compres

Question

Can anyone suggest a way in python to do logging with:

• log rotation every day
• compression of logs when they're rotated
• optional - delete oldest log file to preserve X MB of free space
• optional - sftp log files to server

Thanks for any responses, Fred

Answer 1

I guess it's too late to join the party, but here is what I did. I created a new class inheriting logging.handlers.RotatingFileHandler class and added a couple of lines to gzip the file before moving it.

https://github.com/rkreddy46/python_code_reference/blob/master/compressed_log_rotator.py

#!/usr/bin/env python

# Import all the needed modules
import logging.handlers
import sys
import time
import gzip
import os
import shutil
import random
import string

__version__ = 1.0
__descr__ = "This logic is written keeping in mind UNIX/LINUX/OSX platforms only"


# Create a new class that inherits from RotatingFileHandler. This is where we add the new feature to compress the logs
class CompressedRotatingFileHandler(logging.handlers.RotatingFileHandler):
    def doRollover(self):
        """
        Do a rollover, as described in __init__().
        """
        if self.stream:
            self.stream.close()
        if self.backupCount > 0:
            for i in range(self.backupCount - 1, 0, -1):
                sfn = "%s.%d.gz" % (self.baseFilename, i)
                dfn = "%s.%d.gz" % (self.baseFilename, i + 1)
                if os.path.exists(sfn):
                    # print "%s -> %s" % (sfn, dfn)
                    if os.path.exists(dfn):
                        os.remove(dfn)
                    os.rename(sfn, dfn)
            dfn = self.baseFilename + ".1.gz"
            if os.path.exists(dfn):
                os.remove(dfn)
            # These two lines below are the only new lines. I commented out the os.rename(self.baseFilename, dfn) and
            #  replaced it with these two lines.
            with open(self.baseFilename, 'rb') as f_in, gzip.open(dfn, 'wb') as f_out:
                shutil.copyfileobj(f_in, f_out)
            # os.rename(self.baseFilename, dfn)
            # print "%s -> %s" % (self.baseFilename, dfn)
        self.mode = 'w'
        self.stream = self._open()

# Specify which file will be used for our logs
log_filename = "/Users/myname/Downloads/test_logs/sample_log.txt"

# Create a logger instance and set the facility level
my_logger = logging.getLogger()
my_logger.setLevel(logging.DEBUG)

# Create a handler using our new class that rotates and compresses
file_handler = CompressedRotatingFileHandler(filename=log_filename, maxBytes=1000000, backupCount=10)

# Create a stream handler that shows the same log on the terminal (just for debug purposes)
view_handler = logging.StreamHandler(stream=sys.stdout)

# Add all the handlers to the logging instance
my_logger.addHandler(file_handler)
my_logger.addHandler(view_handler)

# This is optional to beef up the logs
random_huge_data = "".join(random.choice(string.ascii_letters) for _ in xrange(10000))

# All this code is user-specific, write your own code if you want to play around
count = 0
while True:
    my_logger.debug("This is the message number %s" % str(count))
    my_logger.debug(random_huge_data)
    count += 1
    if count % 100 == 0:
        count = 0
        time.sleep(2)

Answer 2

Be warned: The class signatures have changed in python 3. Here is my working example for python 3.6

import logging.handlers
import os
import zlib


def namer(name):
    return name + ".gz"


def rotator(source, dest):
    print(f'compressing {source} -> {dest}')
    with open(source, "rb") as sf:
        data = sf.read()
        compressed = zlib.compress(data, 9)
        with open(dest, "wb") as df:
            df.write(compressed)
    os.remove(source)


err_handler = logging.handlers.TimedRotatingFileHandler('/data/errors.log', when="M", interval=1,
                                                        encoding='utf-8', backupCount=30, utc=True)
err_handler.rotator = rotator
err_handler.namer = namer

logger = logging.getLogger("Rotating Log")
logger.setLevel(logging.ERROR)

logger.addHandler(err_handler)

Answer 3

• log rotation every day: Use a TimedRotatingFileHandler
• compression of logs: Set the encoding='bz2' parameter. (Note this "trick" will only work for Python2. 'bz2' is no longer considered an encoding in Python3.)
• optional - delete oldest log file to preserve X MB of free space. You could (indirectly) arrange this using a RotatingFileHandler. By setting the maxBytes parameter, the log file will rollover when it reaches a certain size. By setting the backupCount parameter, you can control how many rollovers are kept. The two parameters together allow you to control the maximum space consumed by the log files. You could probably subclass the TimeRotatingFileHandler to incorporate this behavior into it as well.

---

Just for fun, here is how you could subclass TimeRotatingFileHandler. When you run the script below, it will write log files to /tmp/log_rotate*.

With a small value for time.sleep (such as 0.1), the log files fill up quickly, reach the maxBytes limit, and are then rolled over.

With a large time.sleep (such as 1.0), the log files fill up slowly, the maxBytes limit is not reached, but they roll over anyway when the timed interval (of 10 seconds) is reached.

All the code below comes from logging/handlers.py. I simply meshed TimeRotatingFileHandler with RotatingFileHandler in the most straight-forward way possible.

import time
import re
import os
import stat
import logging
import logging.handlers as handlers


class SizedTimedRotatingFileHandler(handlers.TimedRotatingFileHandler):
    """
    Handler for logging to a set of files, which switches from one file
    to the next when the current file reaches a certain size, or at certain
    timed intervals
    """

    def __init__(self, filename, maxBytes=0, backupCount=0, encoding=None,
                 delay=0, when='h', interval=1, utc=False):
        handlers.TimedRotatingFileHandler.__init__(
            self, filename, when, interval, backupCount, encoding, delay, utc)
        self.maxBytes = maxBytes

    def shouldRollover(self, record):
        """
        Determine if rollover should occur.

        Basically, see if the supplied record would cause the file to exceed
        the size limit we have.
        """
        if self.stream is None:                 # delay was set...
            self.stream = self._open()
        if self.maxBytes > 0:                   # are we rolling over?
            msg = "%s\n" % self.format(record)
            # due to non-posix-compliant Windows feature
            self.stream.seek(0, 2)
            if self.stream.tell() + len(msg) >= self.maxBytes:
                return 1
        t = int(time.time())
        if t >= self.rolloverAt:
            return 1
        return 0


def demo_SizedTimedRotatingFileHandler():
    log_filename = '/tmp/log_rotate'
    logger = logging.getLogger('MyLogger')
    logger.setLevel(logging.DEBUG)
    handler = SizedTimedRotatingFileHandler(
        log_filename, maxBytes=100, backupCount=5,
        when='s', interval=10,
        # encoding='bz2',  # uncomment for bz2 compression
    )
    logger.addHandler(handler)
    for i in range(10000):
        time.sleep(0.1)
        logger.debug('i=%d' % i)

demo_SizedTimedRotatingFileHandler()

Answer 4

In addition to unutbu's answer: here's how to modify the TimedRotatingFileHandler to compress using zip files.

import logging
import logging.handlers
import zipfile
import codecs
import sys
import os
import time
import glob


class TimedCompressedRotatingFileHandler(logging.handlers.TimedRotatingFileHandler):
    """
    Extended version of TimedRotatingFileHandler that compress logs on rollover.
    """
    def doRollover(self):
        """
        do a rollover; in this case, a date/time stamp is appended to the filename
        when the rollover happens.  However, you want the file to be named for the
        start of the interval, not the current time.  If there is a backup count,
        then we have to get a list of matching filenames, sort them and remove
        the one with the oldest suffix.
        """

        self.stream.close()
        # get the time that this sequence started at and make it a TimeTuple
        t = self.rolloverAt - self.interval
        timeTuple = time.localtime(t)
        dfn = self.baseFilename + "." + time.strftime(self.suffix, timeTuple)
        if os.path.exists(dfn):
            os.remove(dfn)
        os.rename(self.baseFilename, dfn)
        if self.backupCount > 0:
            # find the oldest log file and delete it
            s = glob.glob(self.baseFilename + ".20*")
            if len(s) > self.backupCount:
                s.sort()
                os.remove(s[0])
        #print "%s -> %s" % (self.baseFilename, dfn)
        if self.encoding:
            self.stream = codecs.open(self.baseFilename, 'w', self.encoding)
        else:
            self.stream = open(self.baseFilename, 'w')
        self.rolloverAt = self.rolloverAt + self.interval
        if os.path.exists(dfn + ".zip"):
            os.remove(dfn + ".zip")
        file = zipfile.ZipFile(dfn + ".zip", "w")
        file.write(dfn, os.path.basename(dfn), zipfile.ZIP_DEFLATED)
        file.close()
        os.remove(dfn)

if __name__=='__main__':
    ## Demo of using TimedCompressedRotatingFileHandler() to log every 5 seconds,
    ##     to one uncompressed file and five rotated and compressed files

    os.nice(19)   # I always nice test code

    logHandler = TimedCompressedRotatingFileHandler("mylog", when="S",
        interval=5, backupCount=5) # Total of six rotated log files, rotating every 5 secs
    logFormatter = logging.Formatter(
        fmt='%(asctime)s.%(msecs)03d %(message)s', 
        datefmt='%Y-%m-%d %H:%M:%S'
        )
    logHandler.setFormatter(logFormatter)
    mylogger = logging.getLogger('MyLogRef')
    mylogger.addHandler(logHandler)
    mylogger.setLevel(logging.DEBUG)

    # Write lines non-stop into the logger and rotate every 5 seconds
    ii = 0
    while True:
        mylogger.debug("Test {0}".format(ii))
        ii += 1

Answer 5

The other way to compress logfile during rotate (new in python 3.3) is using BaseRotatingHandler (and all inherited) class attribute rotator for example:

import gzip
import os
import logging
import logging.handlers

class GZipRotator:
    def __call__(self, source, dest):
        os.rename(source, dest)
        f_in = open(dest, 'rb')
        f_out = gzip.open("%s.gz" % dest, 'wb')
        f_out.writelines(f_in)
        f_out.close()
        f_in.close()
        os.remove(dest)

logformatter = logging.Formatter('%(asctime)s;%(levelname)s;%(message)s')
log = logging.handlers.TimedRotatingFileHandler('debug.log', 'midnight', 1, backupCount=5)
log.setLevel(logging.DEBUG)
log.setFormatter(logformatter)
log.rotator = GZipRotator()

logger = logging.getLogger('main')
logger.addHandler(log)    
logger.setLevel(logging.DEBUG)

....

More you can see here.

Answer 6

To copy the file, gzip the copied file (using epoch time), and then clearing out the existing file in a way that won't upset the logging module:

import gzip
import logging
import os
from shutil import copy2
from time import time

def logRoll(logfile_name):
    log_backup_name = logfile_name + '.' + str(int(time()))
    try:
        copy2(logfile_name, log_backup_name)   
    except IOError, err:
        logging.debug(' No logfile to roll')
        return
    f_in = open(log_backup_name, 'rb')
    f_out = gzip.open(log_backup_name + '.gz', 'wb')
    f_out.writelines(f_in)
    f_out.close()
    f_in.close()
    os.remove(log_backup_name)
    f=open(logfile_name, 'w')
    f.close()