Chinaunix首页 | 论坛 | 博客
  • 博客访问: 1366713
  • 博文数量: 245
  • 博客积分: 10021
  • 博客等级: 上将
  • 技术积分: 3094
  • 用 户 组: 普通用户
  • 注册时间: 2008-05-12 14:51
文章存档

2011年(2)

2009年(152)

2008年(91)

我的朋友

分类: LINUX

2009-02-27 14:59:34

SYSKLOGD(8)               Linux System Administration              SYSKLOGD(8)
 
NAME
       sysklogd - Linux system logging utilities.
       linux系统登录功能
 
SYNOPSIS
       syslogd  [ -a socket ] [ -d ] [ -f config file ] [ -h ] [ -l hostlist ]
       [ -m interval ] [ -n ] [ -p socket ] [ -r ] [ -s domainlist ] [ -v ]  [
       -x ]
 
DESCRIPTION
       Sysklogd provides two system utilities which provide support for system
       logging and kernel message trapping.  Support of both internet and unix
       domain  sockets  enables this utility package to support both local and
       remote logging.
       Sysklogd提供了两个系统功用。一个支持系统登录。另一个捕捉内核信息。支持因特网和unix域名套接字能使这项功用支持本地和远端登录。
 
       System logging is provided by a version of syslogd(8) derived from  the
       stock  BSD  sources.   Support  for  kernel  logging is provided by the
       klogd(8) utility which allows kernel logging to be conducted in  either
       a standalone fashion or as a client of syslogd.
       某个版本的syslogd(8)支持记录系统登录是起源于BSD源代码的常备功能。支持内核登录信息的是klogd(8)功用,klogd(8)功用引导内核登录信息到标准样式或者是作为syslogd的客户端。
 
       Syslogd  provides  a  kind  of  logging  that many modern programs use.
       Every logged message contains at least a time  and  a  hostname  field,
       normally  a program name field, too, but that depends on how trusty the
       logging program is.
       Syslogd提供一种记录登录的机制,这种机制被许多现代的程序所使用。 每一个登录信息包含至少登录时间和登录的主机名字。正常情况下,也记录程序的名字,但是这要依赖程序的可信任程度。
 
       While the syslogd sources have been heavily modified a couple of  notes
       are  in  order.   First  of  all there has been a systematic attempt to
       insure that syslogd follows its default, standard  BSD  behavior.   The
       second important concept to note is that this version of syslogd interacts transparently with the version of syslog  found  in  the  standard
       libraries.   If  a binary linked to the standard shared libraries fails
       to function correctly we would like an example of the anomalous  behav-
       ior.
       当syslogd的源代码被严重的修改过后,应该按顺序的注意几个地方。首先系统尝试确保syslogd的默认设置,标准的BSD行为。其次,一个重要的概念是,注意这个版本的syslogd要和在标准库里的其他syslogd配合,找到标准库。如果一个二进制文件连接到一个标准共享库未能保证功能正确的运行。我们希望得到一个不规则运行的例子。
 
       The  main  configuration  file /etc/syslog.conf or an alternative file,
       given with the -f option, is read at startup.   Any  lines  that  begin
       with  the  hash  mark (‘‘#’’) and empty lines are ignored.  If an er
ror occurs during parsing the whole line is ignored.
       主配置文件是/etc/syslog.conf,或是一个非常规的文件,程序开始的时候使用-f参数。忽略以#开头的行和空白行,如果发生错误,忽略全部分析的行。
     
 OPTIONS
       -a socket
              Using this argument you can specify additional sockets from that
              syslogd has to listen to.  This is needed if you’re going to let
              some daemon run within a chroot() environment.  You can  use  up
              to  19 additional sockets.  If your environment needs even more,
              you have to increase the symbol MAXFUNIX  within  the  syslogd.c
              source  file.   An example for a chroot() daemon is described by
              the          people          from           OpenBSD           at
             
              使用这个参数,你可以为syslogd定义额外的套接字来进行端口监听。如果你让一些服务运行在一个chroot环境里,这是必须的。你可以使用19个额外的套接字,如果你的环境需要更多的端口,你必须在源文件里增大symbol MAXFUNIX。
             
 
       -d     Turns  on  debug mode.  Using this the daemon will not proceed a
              fork(2) to set itself in the background, but  opposite  to  that
              stay  in  the foreground and write much debug information on the
              current tty.  See the DEBUGGING section for more information.
 
       -f config file
              Specify an alternative configuration file instead  of  /etc/sys-
              log.conf, which is the default.
 
       -h     By  default  syslogd  will not forward messages it receives from
              remote hosts.  Specifying this switch on the command  line  will
              cause  the log daemon to forward any remote messages it receives
              to forwarding hosts which have been defined.
 
       -l hostlist
              Specify a hostname that should be logged only  with  its  simple
              hostname  and  not  the  fqdn.   Multiple hosts may be specified
              using the colon (‘‘:’’) separator.
 
       -m interval
              The syslogd logs a mark timestamp regularly.  The default inter-
              val  between  two  --  MARK -- lines is 20 minutes.  This can be
              changed with this option.  Setting the interval to zero turns it
              off entirely.
 
       -n     Avoid auto-backgrounding.  This is needed especially if the sys-
              logd is started and controlled by init(8).
 
       -p socket
              You can specify an alternative unix  domain  socket  instead  of
              /dev/log.
 
       -r     This option will enable the facility to receive message from the
              network using an internet domain socket with the syslog  service
              (see  services(5)).   The default is to not receive any messages
              from the network.
 
              This option is introduced in version 1.3 of the  sysklogd  pack-
              age.   Please  note that the default behavior is the opposite of
              how older versions behave, so you might have to turn this on.
 
       -s domainlist
              Specify a domainname that should be stripped off before logging.
              Multiple  domains may be specified using the colon (‘‘:’’) se
pa-
              rator.  Please be advised that no sub-domains may  be  specified
              but  only  entire domains.  For example if -s north.de is speci-
              fied and the host logging resolves to satu.infodrom.north.de  no
              domain  would be cut, you will have to specify two domains like:
              -s north.de:infodrom.north.de.
 
       -v     Print version and exit.
 
       -x     Disable name  lookups  when  receiving  remote  messages.   This
              avoids  deadlocks  when  the  nameserver  is running on the same
              machine that runs the syslog daemon.


SIGNALS
       Syslogd reacts to a set of signals.  You may easily send  a  signal  to
       syslogd using the following:
 
              kill -SIGNAL ‘cat /var/run/syslogd.pid‘
 
       SIGHUP This  lets  syslogd perform a re-initialization.  All open files
              are closed, the configuration file (default is /etc/syslog.conf)
              will be reread and the syslog(3) facility is started again.
              将使syslogd执行重新初始化的行为,关闭所有打开的文件,重新读配置文件,syslog(3)重新开始。
       SIGTERM
              The syslogd will die.
 
       SIGINT, SIGQUIT
              If  debugging  is  enabled  these are ignored, otherwise syslogd
              will die.
 
       SIGUSR1
              Switch debugging on/off.  This option can only be used  if  sys-
              logd is started with the -d debug option.
 
       SIGCHLD
              Wait for childs if some were born, because of wall’ing messages.

CONFIGURATION FILE SYNTAX DIFFERENCES  文件构造语句的差别

       Syslogd uses a slightly different syntax  for  its  configuration  file
       than  the  original BSD sources.  Originally all messages of a specific
       priority and above were forwarded to the log file.

       syslogd的配置文件和BSD源代码的配置文件有些微的差别。默认的,所有指定的优先权,并包括高于此优先权的信息传送到log文件里去。
 
              For example the following line caused ALL  output  from  daemons
              using  the  daemon  facilities (debug is the lowest priority, so
              every higher will also match) to go into /usr/adm/daemons:
              举个例子来说,下列行能让所有使用daemons设备的daemons输出。(debug是最低级的优先权,所以每一个比debug高的优先权将被整合。)输出到/usr/adm/daemons里去。
 
                   # Sample syslog.conf
                   daemon.debug             /usr/adm/daemons
 
       Under the new scheme this behavior remains the same.  The difference is
       the  addition  of  four  new specifiers, the asterisk (*) wildcard, the
       equation sign (=), the exclamation mark (!), and the minus sign (-).
       在新的规划下,这种行为保留各个级别中重复的信息。有区别的时候是使用四个定义符号的时候,这四个定义符号是:通配符星号*,等号=,感叹号!,和减号-。
 
       The * specifies that all messages for the specified facility are to  be
       directed  to  the  destination.   Note that this behavior is degenerate
       with specifying a priority level of debug.  Users have  indicated  that
       the asterisk notation is more intuitive.
       星号*定义了所有指定设备的信息,并且传送到目的地。注意这种行为对定义debug级别的优先权是一种退步。但是用户指出使用星号更加直观。
 
       The  =  wildcard  is used to restrict logging to the specified priority
       class.  This allows, for example, routing only debug messages to a par-
       ticular logging source.
       等号通配符用来表示指定的优先权,比如,允许只传送debug优先权的信息到特定的log文件。
 
              For example the following line in syslog.conf would direct debug
              messages from all sources to the /usr/adm/debug file.
              比如在syslog.conf中的下面这一行将从所有的来源中传送debug信息到/usr/adm/debug文件中。
                   # Sample syslog.conf
                   *.=debug            /usr/adm/debug
 
       The ! is used to exclude logging of  the  specified  priorities.   This
       affects all (!) possibilities of specifying priorities.
       感叹号!排除指定优先权的log信息。会影响到所有可能的定义优先权。
 
              For  example  the  following lines would log all messages of the
              facility mail  except  those  with  the  priority  info  to  the
              /usr/adm/mail file.  And all messages from news.info (including)
              to news.crit (excluding) would be logged  to  the  /usr/adm/news
              file.
              举个例子来说,下面这几行将记录所有除info之外的mail设备的信息到/usr/adm/mail文件里。第二行,将记录所有从news.info(包括info级别)到news.crit(不包括crit)的信息,这些信息记录在/usr/adm/news里。
                   # Sample syslog.conf
                   mail.*;mail.!=info       /usr/adm/mail
                   news.info;news.!crit     /usr/adm/news
 
       You  may  use it intuitively as an exception specifier.  The above men-
       tioned interpretation is simply inverted.  Doing that you may use
       你可以直观的使用排除法的定义,就是上面提及的简单的逆转的意思,要这样做你可以使用
            mail.none
       或者
            mail.!*
       或者
            mail.!debug
      
       to skip every message that comes with a mail facility.  There  is  much
       room to play with it. :-)
       来跳过来自mail设备的每一个信息。有足够的空间来这样做。

       The  -  may  only  be  used  to  prefix  a filename if you want to omit
       sync’ing the file after every write to it.
       -用在文件的前缀中,如果你想在每一次写完文件后,忽略文件的同步,你可以使用 - 。

       日志文件记录设备和服务的信息,但是记录的信息却不是立即写到硬盘中去,使用-,可以使信息先写到log文件中,暂时不写到硬盘中去。
 
       This may take some acclimatization for those individuals  used  to  the
       pure  BSD behavior but testers have indicated that this syntax is some-
       what more flexible than the BSD  behavior.   Note  that  these  changes
       should not affect standard syslog.conf(5) files.  You must specifically
       modify the configuration files to obtain the enhanced behavior.
       这可以解决为单独的文件提供环境适应性。单独的文件被单纯BSD的行为使用,但是试验者指出这个语句比BSD行为有点弹性。注意这些变化不会影响到标准的syslog.conf(5)文件。必须特别修改配置文件以获得增强的行为。
      
 
SUPPORT FOR REMOTE LOGGING 对远端登录的支持。

       These modifications provide network support to  the  syslogd  facility.
       Network support means that messages can be forwarded from one node run-
       ning syslogd to another node running syslogd where they will  be  actu-
       ally logged to a disk file.
       这项修改为syslogd设备提供网络支持。网络支持意味着信息能从一个节点的syslogd传送到另一个正在运行的syslogd节点上,另一个节点的syslogd将实际将登录信息记录到硬盘上。      
 
       To  enable  this you have to specify the -r option on the command line.
       The default behavior is that syslogd won’t listen to the network.
       使用这项功能你必须在命令行里使用-r参数,默认的行为是syslogd不开通网络。
 
       The strategy is to have syslogd listen on  a  unix  domain  socket  for
       locally  generated  log  messages.  This behavior will allow syslogd to
       inter-operate with the syslog found in the standard C library.  At  the
       same time syslogd listens on the standard syslog port for messages for-
       warded from other hosts.  To have this work correctly  the  services(5)
       files (typically found in /etc) must have the following entry:

       这个策略让syslogd监听另一个unix主机域名套接字,为本地产生log信息。这个行为允许syslogd使用在c函数库里的syslog在内部操作。同时syslogd监听标准syslog端口产生的信息,这些信息来自于其他主机。为使这项工作正常进行,services文件里必须有下面这一行。
 
                   syslog          514/udp

 
       If  this  entry  is missing syslogd neither can receive remote messages
       nor send them, because the UDP port cant be  opened.   Instead  syslogd
       will die immediately, blowing out an error message.

       如果这一行丢失,syslogd将不能接受来自远端的信息,也不能发送信息,因为UDP端口没有打开。替代syslogd的程序将消失,错误信息不再保留。
 
       To  cause  messages  to be forwarded to another host replace the normal
       file line in the syslog.conf file with the name of the  host  to  which
       the messages is to be sent prepended with an @.
       要让信息发送到另一台主机,来替代在syslog.conf中定义的正常文件行,通过主机的
名字来实现 ,使用@符号来决定哪些信息被发送了。
              For  example,  to  forward ALL messages to a remote host use the
              following syslog.conf entry:
              举例来说,发送所有信息到远端主机使用下面的配置文件格式。
 
                   # Sample syslogd configuration file to
                   # messages to a remote host forward all.
                   *.*            @hostname
 
              To forward all kernel messages to a remote host  the  configura-
              tion file would be as follows:
              发送所有内核信息到远端主机,在配置文件中应该这样写:
 
                   # Sample configuration file to forward all kernel
                   # messages to a remote host.
                   kern.*         @hostname
 
       If the remote hostname cannot be resolved at startup, because the name-
       server might not be accessible (it may be started  after  syslogd)  you
       don’t  have to worry.  Syslogd will retry to resolve the name ten times
       and then complain.  Another possibility to avoid this is to  place  the
       hostname in /etc/hosts.
       如果远端主机名字在启动的时候未被解析,是因为没有开启域名服务。(可能域名服务在syslogd服务之后开启)。不要担心,Syslogd将重新解析域名十次才结束。另一个避免此种情况的方法是在/etc/hosts里放置主机名。
 
       With  normal  syslogds  you would get syslog-loops if you send out mes-
       sages that were received from a remote host to the same host  (or  more
       complicated to a third host that sends it back to the first one, and so
       on).  In my domain (Infodrom Oldenburg) we accidently got one  and  our
       disks filled up with the same single message. :-(
       通过正常的syslogd,如果你发送出的信息来自于远端主机接受的信息,你可以获得syslog-loops。(或者你接受的信息来自于更复杂一些的第三方主机,这第三方主机向第一个主机发回信息。)。在本地的主机域名里,我们偶尔可以充满一个或四个磁盘的信息。
 
       To  avoid  this  in further times no messages that were received from a
       remote host are sent out to another (or the same) remote host  anymore.
       If  there  are  scenarios where this doesn’t make sense, please drop me
       (Joey) a line.
       为了避免在以后没有信息,从远端主机接受的信息再被发送到另一个远端主机,如果这样规划是不明智的,那么删除一行。
 
       If the remote host is located in the same domain as the  host,  syslogd
       is  running  on, only the simple hostname will be logged instead of the
       whole fqdn.
       如果远端主机在同一个域名系统里,syslogd就进行工作,只是记录相同的主机名,以代替全部的正式域名。
 
       In a local network you may provide a central log server to have all the
       important  information kept on one machine.  If the network consists of
       different domains you don’t have to complain about logging fully quali-
       fied names instead of simple hostnames.  You may want to use the strip-
       domain feature -s of this server.  You can tell the  syslogd  to  strip
       off  several  domains  other  than the one the server is located in and
       only log simple hostnames.
       在本地网络上,你可以准备一个中心log服务器以便让所有的重要信息都保留在一台机器上。如果本地网络包含有不同的域名,你没必要抱怨登录是全部有限制的名字来代替简单的主机名。你可能想使用服务器的域名剥离功能-s。你可以告诉syslogd剥离比一个多一些的其他一系列的域名,在本地只记录一下简单的主机名。
       Using the -l option there’s also a possibility to define  single  hosts
       as  local  machines.   This,  too, results in logging only their simple
       hostnames and not the fqdns.
       使用-l选项也是一种可能,来定义简单的主机和本地机器。这也是在登录里解析简单的主机名,而不是在解析正式域名。
 
       The UDP socket used to forward messages to remote hosts or  to  receive
       messages from them is only opened when it is needed.  In releases prior
       to 1.3-23 it was opened every time but not opened for reading  or  for-
       warding respectively.
       UDP套接字被用来传送信息到远端主机,或者从远端主机接受信息,在需要的时候打开套接字,在较早发行的版本1.3-23里,每次都打开套接字,但是不仅仅为读和传送打开。

OUTPUT TO NAMED PIPES (FIFOs) 输出到管道。

       This  version  of syslogd has support for logging output to named pipes
       (fifos).  A fifo or named pipe can be used as  a  destination  for  log
       messages  by  prepending a pipy symbol (‘‘|’’) to the name of the file.
       This is handy for debugging.  Note that the fifo must be  created  with
       the mkfifo command before syslogd is started.
       本版本的syslogd支持将登录信息输出到命名管道(fifos)。通过为文件的名字预先定义管道符号(‘‘|’’)可以将一个fifo或者是一个命名管道作为登录信息的目的地,这为测试syslogd提供了方便。注意fifo必须在syslogd开启之前,在mkfifo命令行里定义。
 
              The  following configuration file routes debug messages from the
              kernel to a fifo:
              下面的配置文件将从内核里传送出的debug信息传送到fifo里。
 
                   # Sample configuration to route kernel debugging
                   # messages ONLY to /usr/adm/debug which is a
                   # named pipe.
                   kern.=debug              |/usr/adm/debug

INSTALLATION CONCERNS  设置重点

       There is probably one important consideration when installing this ver-
       sion  of  syslogd.  This version of syslogd is dependent on proper for-
       matting of messages by the syslog function.   The  functioning  of  the
       syslog function in the shared libraries changed somewhere in the region
       of libc.so.4.[2-4].n.  The specific change was  to  null-terminate  the
       message  before  transmitting  it to the /dev/log socket.  Proper func-
       tioning of this version of syslogd is dependent on null-termination  of
       the message.
       本节讲述安装这个版本的syslogd设置的重点,这个版本的syslogd依赖由syslog功能产生的有恰当的格式信息。在共享库里的syslog功能改变了libc.so.4.[2-4].n.文件的一些区域。明确的改变是传送信息给/dev/log套接字之前对null-terminate的改变。这个版本的syslogd严格意义上的功能依赖于信息的null-terminate。
      
 
       This  problem  will  typically manifest itself if old statically linked
       binaries are being used on the system.  Binaries using old versions  of
       the syslog function will cause empty lines to be logged followed by the
       message with the first character in  the  message  removed.   Relinking
       these  binaries  to newer versions of the shared libraries will correct
       this problem.
       如果静态链接库使用在系统上,这个程序的代表性明白的是她自己。二进制使用旧版本的syslog功能将导致空行通过跟随的信息被登录,跟随信息的第一个字符被删除。重新连接这些二进制文件到共享库的新版本将正确的执行。

       Both the syslogd(8) and the klogd(8) can either be run from init(8)  or
       started  as part of the rc.*  sequence.  If it is started from init the
       option -n must be set, otherwise you’ll  get  tons  of  syslog  daemons
       started.  This is because init(8) depends on the process ID.
       syslogd(8)和klogd(8)都被init(8)或者是rc.* 级别中的一个级别所启动。如果从init启动,则-n选项必须被设置,否则,大量的syslog守护进程开启。这是因为init(8)依赖于进程ID。
      

SECURITY THREATS  安全威胁
       There  is  the potential for the syslogd daemon to be used as a conduit
       for a denial of service attack.  Thanks  go  to  John  Morrison  (jmor-
       riso@rflab.ee.ubc.ca)  for alerting me to this potential.  A rogue pro-
       gram(mer) could very easily flood the syslogd daemon with  syslog  mes-
       sages  resulting  in the log files consuming all the remaining space on
       the filesystem.  Activating logging over the inet domain  sockets  will
       of  course  expose a system to risks outside of programs or individuals
       on the local machine.
       守护进程可作为拒绝服务攻击的一个管道,syslog服务潜在的有这种危险。感谢你发信息到jmorriso@rflab.ee.ubc.ca来报告这种危险。一个调皮捣蛋的人可以使用syslog信息非常容易的搞鬼。域名套接字当然可以暴露系统的危险。
 
       There are a number of methods of protecting a machine:
 
       1.     Implement kernel firewalling to limit which  hosts  or  networks
              have access to the 514/UDP socket.
 
       2.     Logging  can  be  directed to an isolated or non-root filesystem
              which, if filled, will not impair the machine.
 
       3.     The ext2 filesystem can be used which can be configured to limit
              a  certain  percentage  of  a  filesystem to usage by root only.
              NOTE that this will require syslogd to be run as a non-root pro-
              cess.   ALSO NOTE that this will prevent usage of remote logging
              since syslogd will be unable to bind to the 514/UDP socket.
 
       4.     Disabling inet domain sockets  will  limit  risk  to  the  local
              machine.
 
       5.     Use step 4 and if the problem persists and is not secondary to a
              rogue program/daemon get a 3.5 ft (approx. 1  meter)  length  of
              sucker rod* and have a chat with the user in question.
 
              Sucker  rod  def.  —  3/4,  7/8 or 1in. hardened steel rod, male
              threaded on each end.  Primary use in the oil industry in  West-
              ern North Dakota and other locations to pump ’suck’ oil from oi
l
              wells.  Secondary uses are for the construction of  cattle  feed
              lots  and  for  dealing with the occasional recalcitrant or bel-
              ligerent individual.


 DEBUGGING
       When debugging is turned on using -d option then syslogd will  be  very
       verbose  by  writing much of what it does on stdout.  Whenever the con-
       figuration file is reread and re-parsed you’ll see  a  tabular,  corre-
       sponding to the internal data structure.  This tabular consists of four
       fields:
 
       number This field contains a serial number starting by zero.  This num-
              ber represents the position in the internal data structure (i.e.
              the array).  If one number is left out then there  might  be  an
              error in the corresponding line in /etc/syslog.conf.
 
       pattern
              This  field  is  tricky  and  represents  the internal structure
              exactly.  Every column stands for  a  facility  (refer  to  sys-
              log(3)).   As  you can see, there are still some facilities left
              free for former use, only the left most are used.   Every  field
              in a column represents the priorities (refer to syslog(3)).
 
       action This  field  describes  the  particular  action that takes place
              whenever a message is received that matches the pattern.   Refer
              to the syslog.conf(5) manpage for all possible actions.
 
       arguments
              This field shows additional arguments to the actions in the last
              field.  For file-logging this is the filename for  the  logfile;
              for  user-logging  this  is  a list of users; for remote logging
              this is the hostname of the machine to log to; for  console-log-
              ging this is the used console; for tty-logging this is the spec-
              ified tty; wall has no additional arguments.


阅读(14748) | 评论(0) | 转发(1) |
给主人留下些什么吧!~~