File: README.1st

package info (click to toggle)
sysklogd 1.5-5
  • links: PTS
  • area: main
  • in suites: lenny
  • size: 492 kB
  • ctags: 278
  • sloc: ansic: 3,788; sh: 329; perl: 194; makefile: 185
file content (95 lines) | stat: -rw-r--r-- 4,251 bytes parent folder | download | duplicates (2)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
Important information
---------------------

The included version of syslogd behaves in a slightly different manner
to the one in former releases.  Please review the following important
differences:

* By default the syslog daemon doesn't accept any message from the
  syslog/udp port. To enable this add "-r" to the command-line
  arguments. You _have to_ add this on every host that should run as a
  centralized network log server.

  You also should take a look at other new command-line arguments:
  "-l" and "-s".

  The syslog daemon by default does not forward to a remote host any
  log messages which it received from a remote host.  This is an
  attempt to prevent syslog loops.  If you desire this behavior the
  -h command-line switch can be used to enable this behavior.

* Syslogd was designed to strip off the local domain from every
  message that comes from any host in the same domain. Unfortunately
  this feature didn't work in every cases. This is now corrected. So
  you might not get the fqdn anymore.

  If you use any scripts that analyze logfiles, please bare this in
  mind.

* Syslogd doesn't touch the filemode of any logfile. If it has to
  create one, it will be world-readable. If you do not want this
  (i.e. if auth.* is logged) you have to create the file by hand and
  change permissions.

* If you notice that some of your programs make heavy use of the
  syslog facility and your disks get loud you might want to turn
  fsync()ing after each line off. But note that in doing so you
  increase the likelihood of lost information in the event of a
  system crash.

* If you're going to start klogd or syslogd by init(8), you don't have
  to hack the code anymore. Instead add "-n" to the command-line
  arguments.

* Klogd can now be used to decode EIP addresses if it can determine a
  System.map file (command-line argument "-k"). This is a very useful
  feature if your system crashes, but its usability depends on an
  actual and correct System.map file.

* Both system utilities now check for and respect the existence of .pid
  files.  If the utilities are started by configuration scripts on
  stable systems there is the potential that the process ID numbers of
  the utilities will be identical from one system boot to another.
  This will cause both klogd and syslogd to terminate.

  Both klogd and syslogd will attempt to remove their .pid files when
  they receive termination signals.  The difficulty noted above
  typically occurs when a system crash occurs or an uncatchable signal
  (kill -9) is used to stop the daemons.

  The cleanest solution to this problem is to insure that the system
  configuration scripts (rc.*) provide a clean working environment for
  a freshly booted system.  As part of the initialization process
  these scripts should remove all old .pid files found in /var/run.
  This will insure that klogd and syslogd start properly even if prior
  executions have been terminated harshly.

* Large file support, i.e. support to write to log files that are
  larger than 2 GB is not part of syslogd, but a matter of the glibc
  emitting different system calls to the kernel interface.  To support
  large files you'll have to compile syslogd with the compiler defines
  commented out in the main Makefile so that glibc adjusts the system
  calls.

Modifying the kernel console log level
--------------------------------------

  There are two ways to alter the kernel console log level.  This
  setting controls whether log messages from the kernel should appear
  on the system console or not.

  In the past, klogd had to do this with the -c parameter.  Using '-c
  4' will set the log level of console messages to 4 and only display
  warnings and errors but not regular debug or information messages.

  This behaviour is deprecated and hencely not enforced anymore via
  the RC script of klogd.  Instead sysctl(8) should be used as
  interface to various kernel variables.  These can be stored
  non-volatile in /etc/sysctl.conf.

  The prevent the kernel to flood the system console and to achieve
  the same behaviour of '-c 4' simply add the following to the
  configuration file and let sysctl set this kernel parameter upon
  system boot.

  kernel/printk = 4 4 1 7