check_log.sh 5.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217
  1. #!/bin/sh
  2. #
  3. # Log file pattern detector plugin for Nagios
  4. # Written by Ethan Galstad (nagios@nagios.org)
  5. # Last Modified: 07-31-1999
  6. #
  7. # Usage: ./check_log <log_file> <old_log_file> <pattern>
  8. #
  9. # Description:
  10. #
  11. # This plugin will scan a log file (specified by the <log_file> option)
  12. # for a specific pattern (specified by the <pattern> option). Successive
  13. # calls to the plugin script will only report *new* pattern matches in the
  14. # log file, since an copy of the log file from the previous run is saved
  15. # to <old_log_file>.
  16. #
  17. # Output:
  18. #
  19. # On the first run of the plugin, it will return an OK state with a message
  20. # of "Log check data initialized". On successive runs, it will return an OK
  21. # state if *no* pattern matches have been found in the *difference* between the
  22. # log file and the older copy of the log file. If the plugin detects any
  23. # pattern matches in the log diff, it will return a CRITICAL state and print
  24. # out a message is the following format: "(x) last_match", where "x" is the
  25. # total number of pattern matches found in the file and "last_match" is the
  26. # last entry in the log file which matches the pattern.
  27. #
  28. # Notes:
  29. #
  30. # If you use this plugin make sure to keep the following in mind:
  31. #
  32. # 1. The "max_attempts" value for the service should be 1, as this
  33. # will prevent Nagios from retrying the service check (the
  34. # next time the check is run it will not produce the same results).
  35. #
  36. # 2. The "notify_recovery" value for the service should be 0, so that
  37. # Nagios does not notify you of "recoveries" for the check. Since
  38. # pattern matches in the log file will only be reported once and not
  39. # the next time, there will always be "recoveries" for the service, even
  40. # though recoveries really don't apply to this type of check.
  41. #
  42. # 3. You *must* supply a different <old_file_log> for each service that
  43. # you define to use this plugin script - even if the different services
  44. # check the same <log_file> for pattern matches. This is necessary
  45. # because of the way the script operates.
  46. #
  47. # Examples:
  48. #
  49. # Check for login failures in the syslog...
  50. #
  51. # check_log /var/log/messages ./check_log.badlogins.old "LOGIN FAILURE"
  52. #
  53. # Check for port scan alerts generated by Psionic's PortSentry software...
  54. #
  55. # check_log /var/log/message ./check_log.portscan.old "attackalert"
  56. #
  57. # Paths to commands used in this script. These
  58. # may have to be modified to match your system setup.
  59. GREP="/bin/egrep"
  60. DIFF="/bin/diff"
  61. TAIL="/bin/tail"
  62. CAT="/bin/cat"
  63. RM="/bin/rm"
  64. CHMOD="/bin/chmod"
  65. TOUCH="/bin/touch"
  66. PROGNAME=`/bin/basename $0`
  67. PROGPATH=`echo $0 | sed -e 's,[\\/][^\\/][^\\/]*$,,'`
  68. REVISION="@NP_VERSION@"
  69. PATH="@trusted_path@"
  70. . $PROGPATH/utils.sh
  71. print_usage() {
  72. echo "Usage: $PROGNAME -F logfile -O oldlog -q query"
  73. echo "Usage: $PROGNAME --help"
  74. echo "Usage: $PROGNAME --version"
  75. }
  76. print_help() {
  77. print_revision $PROGNAME $REVISION
  78. echo ""
  79. print_usage
  80. echo ""
  81. echo "Log file pattern detector plugin for Nagios"
  82. echo ""
  83. support
  84. }
  85. # Make sure the correct number of command line
  86. # arguments have been supplied
  87. if [ $# -lt 1 ]; then
  88. print_usage
  89. exit $STATE_UNKNOWN
  90. fi
  91. # Grab the command line arguments
  92. #logfile=$1
  93. #oldlog=$2
  94. #query=$3
  95. exitstatus=$STATE_WARNING #default
  96. while test -n "$1"; do
  97. case "$1" in
  98. --help)
  99. print_help
  100. exit $STATE_OK
  101. ;;
  102. -h)
  103. print_help
  104. exit $STATE_OK
  105. ;;
  106. --version)
  107. print_revision $PROGNAME $REVISION
  108. exit $STATE_OK
  109. ;;
  110. -V)
  111. print_revision $PROGNAME $REVISION
  112. exit $STATE_OK
  113. ;;
  114. --filename)
  115. logfile=$2
  116. shift
  117. ;;
  118. -F)
  119. logfile=$2
  120. shift
  121. ;;
  122. --oldlog)
  123. oldlog=$2
  124. shift
  125. ;;
  126. -O)
  127. oldlog=$2
  128. shift
  129. ;;
  130. --query)
  131. query=$2
  132. shift
  133. ;;
  134. -q)
  135. query=$2
  136. shift
  137. ;;
  138. -x)
  139. exitstatus=$2
  140. shift
  141. ;;
  142. --exitstatus)
  143. exitstatus=$2
  144. shift
  145. ;;
  146. *)
  147. echo "Unknown argument: $1"
  148. print_usage
  149. exit $STATE_UNKNOWN
  150. ;;
  151. esac
  152. shift
  153. done
  154. # If the source log file doesn't exist, exit
  155. if [ ! -e $logfile ]; then
  156. echo "Log check error: Log file $logfile does not exist!"
  157. exit $STATE_UNKNOWN
  158. elif [ ! -r $logfile ] ; then
  159. echo "Log check error: Log file $logfile is not readable!"
  160. exit $STATE_UNKNOWN
  161. fi
  162. # If the old log file doesn't exist, this must be the first time
  163. # we're running this test, so copy the original log file over to
  164. # the old diff file and exit
  165. if [ ! -e $oldlog ]; then
  166. $CAT $logfile > $oldlog
  167. echo "Log check data initialized..."
  168. exit $STATE_OK
  169. fi
  170. # The old log file exists, so compare it to the original log now
  171. # The temporary file that the script should use while
  172. # processing the log file.
  173. if [ -x /bin/mktemp ]; then
  174. tempdiff=`/bin/mktemp /tmp/check_log.XXXXXXXXXX`
  175. else
  176. tempdiff=`/bin/date '+%H%M%S'`
  177. tempdiff="/tmp/check_log.${tempdiff}"
  178. $TOUCH $tempdiff
  179. $CHMOD 600 $tempdiff
  180. fi
  181. $DIFF $logfile $oldlog | $GREP -v "^>" > $tempdiff
  182. # Count the number of matching log entries we have
  183. count=`$GREP -c "$query" $tempdiff`
  184. # Get the last matching entry in the diff file
  185. lastentry=`$GREP "$query" $tempdiff | $TAIL -1`
  186. $RM -f $tempdiff
  187. $CAT $logfile > $oldlog
  188. if [ "$count" = "0" ]; then # no matches, exit with no error
  189. echo "Log check ok - 0 pattern matches found"
  190. exitstatus=$STATE_OK
  191. else # Print total matche count and the last entry we found
  192. echo "($count) $lastentry"
  193. exitstatus=$STATE_CRITICAL
  194. fi
  195. exit $exitstatus