|
ACE学习笔记(二):ACE日志系统(2) 来自:http://masterdog.blogchina.com/blog/984024.html 非常感谢,写的很好!
ACE可以将输出重新定向到stderr(缺省)、系统日志、输出流(Output Stream)、甚至是回调函数,在C/S结构
中,ACE也可以实现客户端输出定向到服务器。本节将详细讨论这些内容。
三 ACE的重新定向设置或改变ACE日志输出目标(logging sink)可以通过ACE_Log_Msg的方法open来完成,另外可通过set_flags和
clr_flags来配合实现输出到多个目标。open方法的原型:
/**
* Initialize the ACE logging facility. Supplies the program name
* that is available to each logging message call. Default arguments
* set up logging to STDERR only.
*
* @param prog_name The name of the calling program.
* @param options_flags A bitwise-or of options flags used to set the
* initial behavior and logging sink(s). (see the
* enum above for the valid values).
* @param logger_key The name of ACE_FIFO rendezvous point where the
* local client logger daemon is listening for logging
* messages. Only meaningful if the LOGGER bit is
* set in the @a flags argument.
*/
int open (const ACE_TCHAR *prog_name,
u_long options_flags = ACE_Log_Msg::STDERR,
const ACE_TCHAR *logger_key = 0);
第一个参数prog_name表示程序名称,可以任意设置,可通过ACE_Log_Msg的program_name方法获取,另外
当第二个参数设置中有ACE_Log_Msg::VERBOSE时,会在输出信息前加上前缀,前缀中就包含该program_name。第二个参数可以是:
STDERR Write messages to STDERR
LOGGER Write messages to the local client logger daemon
OSTREAM Write messages to the assigned output stream
MSG_CALLBACK Write messages to the callback object
VERBOSE Prepends program name, timestamp, host name, process ID, and message priority
to each message
VERBOSE_LITE Prepends timestamp and message priority to each message
SILENT Do not print messages at all
SYSLOG Write messages to the system's event log
CUSTOM Write messages to the user-provided back end第三个参数logger key是针对第二个参数为LOGGER(C/S结构)时需要设置的,表示客户端程序端口值,比如:
ACE_DEFAULT_LOGGER_KEY。一旦调用了该方法,其后的输出将重新定位到第二个参数所指定的sink上。
set_flags、clr_flags的原型:// Enable the bits in the logger's options flags.
void set_flags (unsigned long f);// Disable the bits in the logger's options flags.
void clr_flags (unsigned long f);参数含义同open函数的第二个参数options_flags,其中set_flags调用可在现有输出目标基础上增加新的
输出目标(叠加),clr_flags则从中删除指定的输出目标。下面描述一下重定向到不同sink的例子。
1.输出到stderr这个简单:
ACE_LOG_MSG->open (argv[0], ACE_Log_Msg::STDERR);
ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("test\n")));或者:
ACE_LOG_MSG->set_flags (ACE_Log_Msg::STDERR);
ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("test\n")));2.输出到System LoggerACE_LOG_MSG->open(argv[0], ACE_Log_Msg::SYSLOG, ACE_TEXT ("ACE log"));
ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("test\n")));注意重新定向到System Logger不能用set_flags、clr_flags来实现。Windows操作系统下,会输出到系统的
Event Log中,其中Event Source名称为Open函数第三个参数。Unix/Linux下会输出到syslog facility。另
外Windows下LM_STARTUP、LM_SHUTDOWN、LM_TRACE、LM_DEBUG、LM_INFO 会对应EVENTLOG_INFORMATION_TYPE;
LM_NOTICE、LM_WARNING会对应EVENTLOG_WARNING_TYPE、LM_ERROR、LM_CRITICAL、LM_ALERT、LM_EMERGENCY
会对应EVENTLOG_ERROR_TYPE。3.输出到Output Streams下例将日志输出到文件(ofstream):- #include < ace/Log_Msg.h >
- #include < ace/streams.h >int ACE_TMAIN (int, ACE_TCHAR *argv[])
- {
- ACE_OSTREAM_TYPE *output = new ofstream ("test.txt");
- ACE_LOG_MSG->msg_ostream (output, 1);
- ACE_LOG_MSG->set_flags (ACE_Log_Msg::OSTREAM);
- ACE_LOG_MSG->clr_flags (ACE_Log_Msg::STDERR);
- ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("test\n"))); return 0;
- }
复制代码 上例中首先需要通过msg_ostream来assign一个ACE OSTREAM,然后再指定log sink为ACE_Log_Msg::OSTREAM,
指定log sink既可用set_flags也可用open方法,注意上例中set_flags之后紧接着又用clr_flags屏蔽掉先前
指定的ACE_Log_Msg::STDERR(缺省sink),这样就不会同时输出到stderr了。看一下msg_ostream的注释(解释的很明白,不赘述):/**
* delete_stream == 1, forces Log_Msg.h to delete the stream in
* its own ~dtor (assumes control of the stream)
* use only with proper ostream (eg: fstream), not (cout, cerr)
*/
void msg_ostream (ACE_OSTREAM_TYPE *, int delete_ostream);
上例中由于设置了delete_stream == 1,所以不应再显示地调用delete output;来删除ofstream。
全局性的ostream还包括cout、cerr、clog等,这里再举一个输出到cout的例子,其它情况类似: ACE_LOG_MSG->msg_ostream (&cout);
ACE_LOG_MSG->open (argv[0], ACE_Log_Msg::OSTREAM);
ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("test\n")));4.输出到Callback在ACE中,将日志输出重新定向到回调函数其实很简单,只需要四步:(1)定义ACE_Log_Msg_Callback的派生类,实现纯虚函数virtual void log (ACE_Log_Record &log_record).(2)在ACE_Log_Msg中注册该派生类(比如 MyCallback):
Callback *callback = new Callback;
ACE_LOG_MSG->set_flags (ACE_Log_Msg::MSG_CALLBACK);
ACE_LOG_MSG->clr_flags (ACE_Log_Msg::STDERR);
ACE_LOG_MSG->msg_callback (callback);
(3)调用日志输出宏,比如ACE_DEBUG,就会将输出重新定向到MyCallback的log方法中。(4)退出前的清理工作:
ACE_LOG_MSG->clr_flags (ACE_Log_Msg::MSG_CALLBACK);
delete callback;注意到ACE_Log_Msg_Callback的log方法的参数类型ACE_Log_Record, 可以利用其print方法重新定制输出: int print (const ACE_TCHAR host_name[], u_long verbose_flag, FILE *fp);
或者:
int print (const ACE_TCHAR host_name[], u_long verbose_flag, ACE_OSTREAM_TYPE &stream);
第一个参数表示主机名,第二个参数可选ACE_Log_Msg:: VERBOSE、VERBOSE_LITE或0(原始信息),
第三个参数可以是文件指针或者是标准输出流,比如: log_record.print (ACE_TEXT (""), ACE_Log_Msg::VERBOSE, cout); /* 输出到屏幕 */
log_record.print (ACE_TEXT (""), 0, ofstream("test.txt",ios::app)); /* 输出到文件 */关于ACE_Log_Record进一步使用,请参考以下例子:- /* main.cpp */
- #include < ace/Task.h >
- #include < ace/Log_Msg.h >
- #include " MyCallback.h "int ACE_TMAIN (int, ACE_TCHAR *[])
- {
- MyCallback *callback = new MyCallback; ACE_LOG_MSG->set_flags (ACE_Log_Msg::MSG_CALLBACK);
- ACE_LOG_MSG->clr_flags (ACE_Log_Msg::STDERR);
- ACE_LOG_MSG->msg_callback (callback); ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("test1\n")));
- ACE_DEBUG ((LM_INFO, ACE_TEXT ("test2\n"))); ACE_LOG_MSG->clr_flags (ACE_Log_Msg::MSG_CALLBACK);
- delete callback; return 0;
- }
- /* MyCallback.h */
- #include < ace/streams.h >
- #include < ace/Log_Msg.h >
- #include < ace/Log_Msg_Callback.h >
- #include < ace/Log_Record.h >
- #include < ace/streams.h >
- #include < ace/Log_Msg_Callback.h >
- #include < ace/Log_Record.h >
- #include < ace/SString.h >
- #include < ace/OS.h >class MyCallback : public ACE_Log_Msg_Callback
- {
- public:
- void log (ACE_Log_Record &log_record)
- {
- cerr << "Log Message Received:" << endl;
- unsigned long msg_severity = log_record.type (); ACE_Log_Priority prio = ACE_static_cast (ACE_Log_Priority, msg_severity);
- const ACE_TCHAR *prio_name = ACE_Log_Record::priority_name (prio);
- cerr << "\tType: "
- << ACE_TEXT_ALWAYS_CHAR (prio_name)
- << endl; cerr << "\tLength: " << log_record.length () << endl; const time_t epoch = log_record.time_stamp ().sec ();
- cerr << "\tTime_Stamp: "
- << ACE_TEXT_ALWAYS_CHAR (ACE_OS::ctime (&epoch))
- << flush; cerr << "\tPid: " << log_record.pid () << endl; ACE_CString data (">> ");
- data += ACE_TEXT_ALWAYS_CHAR (log_record.msg_data ()); cerr << "\tMsgData: " << data.c_str () << endl;
- }
- };
复制代码 输出:
Log Message Received:
Type: LM_DEBUG
Length: 32
Time_Stamp: Mon Mar 27 17:03:06 2005
Pid: 2752
MsgData: >> test1Log Message Received:
Type: LM_INFO
Length: 32
Time_Stamp: Mon Mar 27 17:03:06 2005
Pid: 2752
MsgData: >> test2
需要提醒的是,由于ACE_Log_Msg是线程相关的,而且spawn出来的线程没有继承性,因此Callback仅对单个
线程有效,如果想实现多个线程的callback,需要分别在各个线程中调用:
ACE_LOG_MSG->msg_callback (callback);5.C/S结构中输出到服务器可用于分布式系统的日志处理(distributed logger),logging server运行在某台主机,并接收来自其它主机
的logging requests. 在其它主机(也可能在本机)需要运行一个logging client daemon进程,相当于一个
proxy,实现了你自己编写的程序与logging server的交互。自己编写的程序格式如下:- #include < ace/Log_Msg.h >
- int ACE_TMAIN (int, ACE_TCHAR *argv[])
- {
- ACE_LOG_MSG->open (argv[0],
- ACE_Log_Msg:OGGER,
- ACE_DEFAULT_LOGGER_KEY); ACE_DEBUG ((LM_DEBUG, ACE_TEXT ("Sent to Logging Server\n"))); return 0;
复制代码 }在运行你自己编写的程序之前需要:
(1)首先运行logging server:
%ACE_ROOT%/netsvcs/servers/main -f server.conf
(2)然后运行logging client daemon:
%ACE_ROOT%/netsvcs/servers/main -f client.conf
其中server.conf内容如下:
dynamic Logger Service_Object * ACE:_make_ACE_Logging_Strategy() "-s log.txt -f STDERR|OSTREAM"
dynamic Server_Logging_Service Service_Object * netsvcs:_make_ACE_Server_Logging_Acceptor()
active "-p 20009"
该文件每行以dynamic开始,第一行定义了logging strategy,将会把日志输出到log.txt(-s 参数),同时
输出到STDERR。
第二行定义了logging server侦听(listen)端口为20009。client.conf内容如下:
dynamic Client_Logging_Service Service_Object * netsvcs:_make_ACE_Client_Logging_Acceptor()
active "-p 20009 -h localhost"
该文件同样以dynamic开始,定义了logging client daemon的侦听端口为2009(与server.conf一致),同时
-h指定目标机器名称或ip(这里是本机)。
运行了logging server和logging client daemon之后,编译运行上面那个程序, logging server会显示:
Sent to Logging Server同时在logging server所在目录(%ACE_ROOT%/netsvcs/servers/)下会生成log.txt,内容是:
starting up Logging Server at port 20009 on handle 1900
Sent to Logging Server
注意到之所以如此麻烦地启动两级logging管理,主要是考虑到分布式处理过程中可能出现的瓶颈问题。当然也可以直接通过ACE_SOCK_Stream实现与logging server的交互,而不用logging client daemon,细节以后讨论。
而利用server.conf和client.conf来配置分布式logging服务涉及到了ACE的Runtime Configuration,
细节同样在以后会讨论到。对其中参数的解释列举如下:-f Specify ACE_Log_Msg flags (OSTREAM, STDERR, LOGGER, VERBOSE, SILENT, VERBOSE_LITE) used to
control logging.
-i The interval, in seconds, at which the log file size is sampled (default is 0; do not sample
by default).
-k Specify the rendezvous point for the client logger(比如ACE_DEFAULT_LOGGER_KEY)
-m The maximum log file size in Kbytes.
-n Set the program name for the %n format specifier.
-N The maximum number of log files to create.
-o Request the standard log file ordering (keeps multiple log files in numbered order). Default
is not to order log files.
-p Pass in the processwide priorities to either enable (DEBUG, INFO, WARNING, NOTICE, ERROR,
CRITICAL, ALERT, EMERGENCY) or to disable (~DEBUG, ~INFO, ~WARNING, ~NOTICE, ~ERROR, ~CRITICAL,
~ALERT, ~EMERGENCY).
-s Specify the file name used when OSTREAM is specified as an output target.
-t Pass in the per instance priorities to either enable (DEBUG, INFO, WARNING, NOTICE, ERROR,
CRITICAL, ALERT, EMERGENCY) or to disable (~DEBUG, ~INFO, ~WARNING, ~NOTICE, ~ERROR, ~CRITICAL,
~ALERT, ~EMERGENCY).
-w Cause the log file to be wiped out on both start-up and reconfiguration.
[ 本帖最后由 peakzhang 于 2007-12-19 00:19 编辑 ] |
|