Django 使用 Python 内置的 logging 模块处理系统日志。关于该模块的使用，Python 文档里有更详细的讨论。不过，如果你从未用过 Python 的 logging 框架（或者即便你用过），这里是一篇快速入门。
一份 Python logging 配置有下面四个部分组成：
\nlogger 是日志系统的入口。每个 logger 都是命名了的 bucket， 消息写入 bucket 以便进一步处理。
\nlogger 可以配置 日志级别。日志级别描述了由该 logger 处理的消息的严重性。Python 定义了下面几种日志级别：
\nDEBUG：排查故障时使用的低级别系统信息
INFO：一般的系统信息
WARNING：描述系统发生了一些小问题的信息
ERROR：描述系统发生了大问题的信息
CRITICAL：描述系统发生严重问题的信息
每一条写入 logger 的消息都是一条 日志记录。每一条日志记录也包含 日志级别，代表对应消息的严重程度。日志记录还包含有用的元数据，来描述被记录了日志的事件细节，例如堆栈跟踪或者错误码。
\n当 logger 处理一条消息时，会将自己的日志级别和这条消息的日志级别做对比。如果消息的日志级别匹配或者高于 logger 的日志级别，它就会被进一步处理。否则这条消息就会被忽略掉。
\n当 logger 确定了一条消息需要处理之后，会把它传给 Handler。
\nHandler 是决定如何处理 logger 中每一条消息的引擎。它描述特定的日志行为，比如把消息输出到屏幕、文件或网络 socket。
\n和 logger 一样，handler 也有日志级别的概念。如果一条日志记录的级别不匹配或者低于 handler 的日志级别，对应的消息会被 handler 忽略。
\n一个 logger 可以有多个 handler，每一个 handler 可以有不同的日志级别。这样就可以根据消息的重要性不同，来提供不同格式的输出。例如，你可以添加一个 handler 把 ERROR 和 CRITICAL 消息发到寻呼机，再添加另一个 handler 把所有的消息（包括 ERROR 和 CRITICAL 消息）保存到文件里以便日后分析。
在日志记录从 logger 传到 handler 的过程中，使用 Filter 来做额外的控制。
\n默认情况下，只要级别匹配，任何日志消息都会被处理。不过，也可以通过添加 filter 来给日志处理的过程增加额外条件。例如，可以添加一个 filter 只允许某个特定来源的 ERROR 消息输出。
Filter 还被用来在日志输出之前对日志记录做修改。例如，可以写一个 filter，当满足一定条件时，把日志记录从 ERROR 降到 WARNING 级别。
Filter 在 logger 和 handler 中都可以添加；多个 filter 可以链接起来使用，来做多重过滤操作。
\n日志记录最终是需要以文本来呈现的。Formatter 描述了文本的格式。一个 formatter 通常由包含 LogRecord attributes 的 Python 格式化字符串组成，不过你也可以为特定的格式来配置自定义的 formatter。
\n配置好了 logger，handler，filter 和 formatter 之后，需要在代码里发起 logging 的调用。使用 logging 框架非常简单，下面是个例子：
\n# import the logging library\nimport logging\n\n# Get an instance of a logger\nlogger = logging.getLogger(__name__)\n\ndef my_view(request, arg1, arg):\n ...\n if bad_mojo:\n # Log an error message\n logger.error('Something went wrong!')\n就这么简单！bad_mojo 条件每次满足都会写一条 error 日志。
对 logging.getLogger() 的调用会获取（必要时会创建）一个 logger 的实例。不同的 logger 实例用名字来区分。这个名字是为了在配置的时候指定 logger。
按照惯例，logger 的名字通常是包含该 logger 的 Python 模块名，即 __name__。这样可以基于模块来过滤和处理日志请求。不过，如果你有其他的方式来组织你的日志消息，可以为 logger 提供点号分割的名字来标识它：
# Get an instance of a specific named logger\nlogger = logging.getLogger('project.interesting.stuff')\n这种 logger 的名字，用点号分隔的路径定义了一种层次结构。project.interesting 这个 logger 是 project.interesting.stuff logger 的上级；而 project logger 是 project.interesting logger 的上级。
为什么这种层级结构是重要的呢？因为 logger 可以设置为将日志的请求*传播*给上级。这样就可以在 logger 树结构的顶层定义一组单独的 handler，来捕获所有下层的日志请求。在 project 命名空间中定义的 logger handler 将会捕获 project.interesting 和 project.interesting.stuff 这两个 logger 中的所有日志请求。
可以基于 logger 来控制传播的行为。 如果你不希望某个 logger 传播给上级，可以关闭它。
\nlogger 实例包含了每种默认日志级别的入口方法：
\nlogger.debug()
logger.info()
logger.warning()
logger.error()
logger.critical()
还有两种其他的调用方法：
\nlogger.log()：手动输出一条指定日志级别的日志消息。
logger.exception()：创建一个包含当前异常堆栈帧的 ERROR 级别日志消息。
当然，仅仅在代码里调用 logging 是不够的。还需要配置 logger、handler、filter 和 formatter 来确保日志框架能有效地输出日志。
\nPython 的日志库提供了一些配置方法，可以使用编程接口或者配置文件。Django默认使用 dictConfig format。
\n为了配置 logging ，用字典的格式定义一个 LOGGING 配置项，这些配置描述了你想要的 logger、handler、filter 和 formatter，以及它们的日志级别和其他你想要的属性。
默认情况下 LOGGING 配置和 Django 默认日志配置 按照下面的方式合并在一起：
If the disable_existing_loggers key in the LOGGING dictConfig is\nset to True (which is the dictConfig default if the key is missing)\nthen all loggers from the default configuration will be disabled. Disabled\nloggers are not the same as removed; the logger will still exist, but will\nsilently discard anything logged to it, not even propagating entries to a\nparent logger. Thus you should be very careful using\n'disable_existing_loggers': True; it's probably not what you want. Instead,\nyou can set disable_existing_loggers to False and redefine some or all\nof the default loggers; or you can set LOGGING_CONFIG to None\nand handle logging config yourself.
logging 被配置成了 Django setup() 函数的一部分。因此，你可以确定的是，logger 一直都可以在项目代码里使用。
dictConfig format 文档是获取日志配置细节的最好资料。不过，为了让你知道能做什么，下面有几个例子。
\n首先，这是一个简单配置，会将所有来自 django 的日志输出至一个本地文件：
\nLOGGING = {\n 'version': 1,\n 'disable_existing_loggers': False,\n 'handlers': {\n 'file': {\n 'level': 'DEBUG',\n 'class': 'logging.FileHandler',\n 'filename': '/path/to/django/debug.log',\n },\n },\n 'loggers': {\n 'django': {\n 'handlers': ['file'],\n 'level': 'DEBUG',\n 'propagate': True,\n },\n },\n}\n若你使用此例子，切记要将 'filename' 指向的路径改为当前运行 Django 应用的用户可写的路径。
接着是一个如何使日志系统将 Django 的日志输出至控制台的例子。这在本地调试时会很实用。
\n默认情况下，该配置只会将 INFO 或以上的信息发送至终端（与 Django 的默认配置相同，除了 Django 仅在 DEBUG=True 时显示日志）。Django 并未打印太多信息。你也可以设置环境变量 DJANGO_LOG_LEVEL=DEBUG ，这样可以看到 Django 的所有调试日志，这将会很冗长，因为它包含了所有的数据库查询记录：
import os\n\nLOGGING = {\n 'version': 1,\n 'disable_existing_loggers': False,\n 'handlers': {\n 'console': {\n 'class': 'logging.StreamHandler',\n },\n },\n 'loggers': {\n 'django': {\n 'handlers': ['console'],\n 'level': os.getenv('DJANGO_LOG_LEVEL', 'INFO'),\n },\n },\n}\n最后是一个完整的日志配置例子：
\nLOGGING = {\n 'version': 1,\n 'disable_existing_loggers': False,\n 'formatters': {\n 'verbose': {\n 'format': '{levelname} {asctime} {module} {process:d} {thread:d} {message}',\n 'style': '{',\n },\n 'simple': {\n 'format': '{levelname} {message}',\n 'style': '{',\n },\n },\n 'filters': {\n 'special': {\n '()': 'project.logging.SpecialFilter',\n 'foo': 'bar',\n },\n 'require_debug_true': {\n '()': 'django.utils.log.RequireDebugTrue',\n },\n },\n 'handlers': {\n 'console': {\n 'level': 'INFO',\n 'filters': ['require_debug_true'],\n 'class': 'logging.StreamHandler',\n 'formatter': 'simple'\n },\n 'mail_admins': {\n 'level': 'ERROR',\n 'class': 'django.utils.log.AdminEmailHandler',\n 'filters': ['special']\n }\n },\n 'loggers': {\n 'django': {\n 'handlers': ['console'],\n 'propagate': True,\n },\n 'django.request': {\n 'handlers': ['mail_admins'],\n 'level': 'ERROR',\n 'propagate': False,\n },\n 'myproject.custom': {\n 'handlers': ['console', 'mail_admins'],\n 'level': 'INFO',\n 'filters': ['special']\n }\n }\n}\n该日志配置做了以下事情：
\nIdentifies the configuration as being in 'dictConfig version 1'\nformat. At present, this is the only dictConfig format version.
Defines two formatters:
\nsimple, that just outputs the log level name (e.g.,\nDEBUG) and the log message.
The format string is a normal Python formatting string\ndescribing the details that are to be output on each logging\nline. The full list of detail that can be output can be\nfound in Formatter Objects.
verbose, that outputs the log level name, the log\nmessage, plus the time, process, thread and module that\ngenerate the log message.
Defines two filters:
\nproject.logging.SpecialFilter, using the alias special. If this\nfilter required additional arguments, they can be provided as additional\nkeys in the filter configuration dictionary. In this case, the argument\nfoo will be given a value of bar when instantiating\nSpecialFilter.
django.utils.log.RequireDebugTrue, which passes on records when\nDEBUG is True.
Defines two handlers:
\nconsole, a StreamHandler, which prints any INFO\n(or higher) message to sys.stderr. This handler uses the simple\noutput format.
mail_admins, an AdminEmailHandler, which emails any ERROR\n(or higher) message to the site ADMINS. This handler uses the\nspecial filter.
Configures three loggers:
\ndjango, which passes all messages to the console handler.
django.request, which passes all ERROR messages to\nthe mail_admins handler. In addition, this logger is\nmarked to not propagate messages. This means that log\nmessages written to django.request will not be handled\nby the django logger.
myproject.custom, which passes all messages at INFO\nor higher that also pass the special filter to two\nhandlers -- the console, and mail_admins. This\nmeans that all INFO level messages (or higher) will be\nprinted to the console; ERROR and CRITICAL\nmessages will also be output via email.
If you don't want to use Python's dictConfig format to configure your\nlogger, you can specify your own configuration scheme.
\nThe LOGGING_CONFIG setting defines the callable that will\nbe used to configure Django's loggers. By default, it points at\nPython's logging.config.dictConfig() function. However, if you want to\nuse a different configuration process, you can use any other callable\nthat takes a single argument. The contents of LOGGING will\nbe provided as the value of that argument when logging is configured.
If you don't want to configure logging at all (or you want to manually\nconfigure logging using your own approach), you can set\nLOGGING_CONFIG to None. This will disable the\nconfiguration process for Django's default logging. Here's an example that disables Django's\nlogging configuration and then manually configures logging:
LOGGING_CONFIG = None\n\nimport logging.config\nlogging.config.dictConfig(...)\nSetting LOGGING_CONFIG to None only means that the automatic\nconfiguration process is disabled, not logging itself. If you disable the\nconfiguration process, Django will still make logging calls, falling back to\nwhatever default logging behavior is defined.
Django provides a number of utilities to handle the unique\nrequirements of logging in Web server environment.
\nDjango provides several built-in loggers.
\ndjangoLink to this heading#The catch-all logger for messages in the django hierarchy. No messages are\nposted using this name but instead using one of the loggers below.
django.requestLink to this heading#Log messages related to the handling of requests. 5XX responses are\nraised as ERROR messages; 4XX responses are raised as WARNING\nmessages. Requests that are logged to the django.security logger aren't\nlogged to django.request.
Messages to this logger have the following extra context:
\nstatus_code: The HTTP response code associated with the\nrequest.
request: The request object that generated the logging\nmessage.
django.serverLink to this heading#Log messages related to the handling of requests received by the server invoked\nby the runserver command. HTTP 5XX responses are logged as ERROR\nmessages, 4XX responses are logged as WARNING messages, and everything else\nis logged as INFO.
Messages to this logger have the following extra context:
\nstatus_code: The HTTP response code associated with the request.
request: The request object that generated the logging message.
django.templateLink to this heading#Log messages related to the rendering of templates.
\nMissing context variables are logged as DEBUG messages.
django.db.backendsLink to this heading#Messages relating to the interaction of code with the database. For example,\nevery application-level SQL statement executed by a request is logged at the\nDEBUG level to this logger.
Messages to this logger have the following extra context:
\nduration: The time taken to execute the SQL statement.
sql: The SQL statement that was executed.
params: The parameters that were used in the SQL call.
For performance reasons, SQL logging is only enabled when\nsettings.DEBUG is set to True, regardless of the logging\nlevel or handlers that are installed.
This logging does not include framework-level initialization (e.g.\nSET TIMEZONE) or transaction management queries (e.g. BEGIN,\nCOMMIT, and ROLLBACK). Turn on query logging in your database if you\nwish to view all database queries.
django.security.*Link to this heading#The security loggers will receive messages on any occurrence of\nSuspiciousOperation and other security-related\nerrors. There is a sub-logger for each subtype of security error, including all\nSuspiciousOperations. The level of the log event depends on where the\nexception is handled. Most occurrences are logged as a warning, while\nany SuspiciousOperation that reaches the WSGI handler will be logged as an\nerror. For example, when an HTTP Host header is included in a request from\na client that does not match ALLOWED_HOSTS, Django will return a 400\nresponse, and an error message will be logged to the\ndjango.security.DisallowedHost logger.
These log events will reach the django logger by default, which mails error\nevents to admins when DEBUG=False. Requests resulting in a 400 response due\nto a SuspiciousOperation will not be logged to the django.request\nlogger, but only to the django.security logger.
To silence a particular type of SuspiciousOperation, you can override that\nspecific logger following this example:
'handlers': {\n 'null': {\n 'class': 'logging.NullHandler',\n },\n},\n'loggers': {\n 'django.security.DisallowedHost': {\n 'handlers': ['null'],\n 'propagate': False,\n },\n},\nOther django.security loggers not based on SuspiciousOperation are:
django.security.csrf: For CSRF failures.
django.db.backends.schemaLink to this heading#Logs the SQL queries that are executed during schema changes to the database by\nthe migrations framework. Note that it won't log the\nqueries executed by RunPython.\nMessages to this logger have params and sql in their extra context (but\nunlike django.db.backends, not duration). The values have the same meaning\nas explained in django.db.backends.
Django provides one log handler in addition to those provided by the\nPython logging module.
\nThis handler sends an email to the site ADMINS for each log\nmessage it receives.
If the log record contains a request attribute, the full details\nof the request will be included in the email. The email subject will\ninclude the phrase "internal IP" if the client's IP address is in the\nINTERNAL_IPS setting; if not, it will include "EXTERNAL IP".
If the log record contains stack trace information, that stack\ntrace will be included in the email.
\nThe include_html argument of AdminEmailHandler is used to\ncontrol whether the traceback email includes an HTML attachment\ncontaining the full content of the debug Web page that would have been\nproduced if DEBUG were True. To set this value in your\nconfiguration, include it in the handler definition for\ndjango.utils.log.AdminEmailHandler, like this:
'handlers': {\n 'mail_admins': {\n 'level': 'ERROR',\n 'class': 'django.utils.log.AdminEmailHandler',\n 'include_html': True,\n }\n},\nNote that this HTML version of the email contains a full traceback,\nwith names and values of local variables at each level of the stack, plus\nthe values of your Django settings. This information is potentially very\nsensitive, and you may not want to send it over email. Consider using\nsomething such as Sentry to get the best of both worlds -- the\nrich information of full tracebacks plus the security of not sending the\ninformation over email. You may also explicitly designate certain\nsensitive information to be filtered out of error reports -- learn more on\nFiltering error reports.
\nBy setting the email_backend argument of AdminEmailHandler, the\nemail backend that is being used by the\nhandler can be overridden, like this:
'handlers': {\n 'mail_admins': {\n 'level': 'ERROR',\n 'class': 'django.utils.log.AdminEmailHandler',\n 'email_backend': 'django.core.mail.backends.filebased.EmailBackend',\n }\n},\nBy default, an instance of the email backend specified in\nEMAIL_BACKEND will be used.
Sends emails to admin users. To customize this behavior, you can\nsubclass the AdminEmailHandler class and\noverride this method.
Django provides some log filters in addition to those provided by the Python\nlogging module.
\nThis filter accepts a callback function (which should accept a single\nargument, the record to be logged), and calls it for each record that\npasses through the filter. Handling of that record will not proceed if the\ncallback returns False.
\nFor instance, to filter out UnreadablePostError\n(raised when a user cancels an upload) from the admin emails, you would\ncreate a filter function:
from django.http import UnreadablePostError\n\ndef skip_unreadable_post(record):\n if record.exc_info:\n exc_type, exc_value = record.exc_info[:2]\n if isinstance(exc_value, UnreadablePostError):\n return False\n return True\nand then add it to your logging config:
\n'filters': {\n 'skip_unreadable_posts': {\n '()': 'django.utils.log.CallbackFilter',\n 'callback': skip_unreadable_post,\n }\n},\n'handlers': {\n 'mail_admins': {\n 'level': 'ERROR',\n 'filters': ['skip_unreadable_posts'],\n 'class': 'django.utils.log.AdminEmailHandler'\n }\n},\nThis filter will only pass on records when settings.DEBUG is False.
\nThis filter is used as follows in the default LOGGING\nconfiguration to ensure that the AdminEmailHandler only sends\nerror emails to admins when DEBUG is False:
'filters': {\n 'require_debug_false': {\n '()': 'django.utils.log.RequireDebugFalse',\n }\n},\n'handlers': {\n 'mail_admins': {\n 'level': 'ERROR',\n 'filters': ['require_debug_false'],\n 'class': 'django.utils.log.AdminEmailHandler'\n }\n},\nThis filter is similar to RequireDebugFalse, except that records are\npassed only when DEBUG is True.
By default, Django configures the following logging:
\nWhen DEBUG is True:
The django logger sends messages in the django hierarchy (except\ndjango.server) at the INFO level or higher to the console.
When DEBUG is False:
The django logger sends messages in the django hierarchy (except\ndjango.server) with ERROR or CRITICAL level to\nAdminEmailHandler.
Independent of the value of DEBUG:
The django.server logger sends messages at the INFO level\nor higher to the console.
All loggers except django.server propagate logging to their\nparents, up to the root django logger. The console and mail_admins\nhandlers are attached to the root logger to provide the behavior described\nabove.
See also Configuring logging to learn how you can\ncomplement or replace this default logging configuration defined in\ndjango/utils/log.py.
\n