Merge c9271972a6 into fc9c1902ab

nagios/check-netapp-ng.pl

@@ -30,8 +30,21 @@
 ## AUTOSUPPORTSTATUS|NFSOPS|
 ## CIFSOPS|SHELFINFO|...
 ##
-#####################################
+##
-#####################################
+##
+##  CLARIFICATION FOR TIMEOUTS
+##  There are multiples timeouts we depend on
+##
+##  Perl plugins timeout (utils.pm)--- > $TIMEOUT --- > originally 15 sec --> recommandation to raise it to 180
+##  
+##  Net::SNMP timeout --- > Used in Net::SNMP->session --- > originally 5 sec --> recommandation to raise it to 60
+##  Beware that the max value 60 seconds. If set above you get the error message "Can't create snmp session"
+##
+##  Do not forget that in nagios you need to increase service_check_timeout to a value above $TIMEOUT_PLUGINS
+##  Nagios service check timeout (nagios.cfg) --- > service_check_timeout=240 --- > originally 30 sec
+##
+####################################################################################################################################################
+####################################################################################################################################################
 
 
 use strict;
@@ -47,13 +60,16 @@ use Getopt::Long;
 use Time::Local;
 use IPC::Cmd qw(run_forked);
 
+
+
 Getopt::Long::Configure('bundling');
 
+my $TIMEOUT_PLUGINS=$TIMEOUT;
 my $stat = 0;
 my $msg;
 my $perf;
 my $script_name = basename($0);
-my $script_version = 1.3;
+my $script_version = 1.3.1;
 
 my $counterFilePath="/tmp";
 my $counterFile;
@@ -339,9 +355,7 @@ sub FSyntaxError($) {
     my $err = shift;
     print <<EOU;
     $err
-
 	This is $script_name in version $script_version.
-
       Syntax:
 	-H <IP_or_Hostname>     Ip/Dns Name of the Filer
 	-C <community_name>     SNMP Community Name for read
@@ -356,7 +370,6 @@ This is $script_name in version $script_version.
 	-e <vol1[,vol2[,...]]>  Exclude volumes from snap check (SNAPSHOT/SNAPSHOTAGE)
 	-I                      Inform only, return OK every time (ignore -w and -c values)
 	-h                      This help
-
 	Available check types:
 	TEMP              - Temperature
 	FAN               - Fan Fail
@@ -382,20 +395,15 @@ This is $script_name in version $script_version.
 	UPTIME            - Only show\'s uptime
 	CACHEAGE          - Cache Age (-w -c)
 	FSSTATUS          - Overall file system health
-
       Examples:
 	$script_name -H netapp.mydomain -C public -T UPTIME
       UPTIME: 2 days, 23:03:21.09 | uptime=255801s
-
       $script_name -H netapp.mydomain -C public -T DISKUSED -v /vol/data/ -w 90 -c 95 -V 2c
     OK: DISKUSED 79% | /vol/data/=8104595240k
-
     $script_name -H netapp.mydomain -C public -T GLOBALSTATUS
   CRIT: GLOBALSTATUS nonCritical 4 Disk on adapter 1a, shelf 1, bay 9, failed.   | globalstatus=4
-
   $script_name -H netapp.mydomain -C public -T DISKUSED -v wtf
  WARN: Unknown volume path or aggregate name 'wtf'. Available values: aggr_p1a_sas2_mirror /vol/vol0/ /vol/esx/ /vol/xen_a/
-
 EOU
  exit($ERRORS{'UNKNOWN'});
 }
@@ -493,6 +501,8 @@ $opt{'crit'} = 500;
 $opt{'warn'} = 500;
 $opt{'version'} = 2;
 $opt{'timeout'} = 60;
+$TIMEOUT_PLUGINS = 180 ;
+
 my $result = GetOptions(\%opt,
 			'filer|H=s',
 			'community|C=s',
@@ -507,6 +517,17 @@ my $result = GetOptions(\%opt,
 			"help|h",
     );
 
+if ( $opt{'timeout'} > 60)
+{
+    #Set timeout for plugin to the parameter received via command line, but set snmp timeout to the max (60 seconds) if CLI timeout is above 60 seconds
+    $TIMEOUT_PLUGINS=$opt{'timeout'};
+    $opt{'timeout'}=60;
+} else
+{
+     $TIMEOUT_PLUGINS=$opt{'timeout'};
+}    
+
+
 FSyntaxError("") if defined $opt{'help'};
 FSyntaxError("Missing -H")  unless defined $opt{'filer'};
 FSyntaxError("Missing -C")  unless defined $opt{'community'};
@@ -536,9 +557,16 @@ if (!defined($counterFilePath)) {
 
 
 
+# Just in case of problems, let's not hang Nagios
+# with "Return code of 142 is out of bounds", instead we set the message "No response in time"
+
+$SIG{'ALRM'} = sub {
+    print ("CRITICAL: No response in time\n");
+    exit $ERRORS{"CRITICAL"};
+};
 
 # Starting Alarm
-alarm($TIMEOUT);
+alarm($TIMEOUT_PLUGINS);
 
 # Establish SNMP Session
 our $snmp_session = _create_session($opt{'filer'},$opt{'community'},$opt{'version'},$opt{'timeout'});
@@ -1149,3 +1177,4 @@ $msg =~ s/\n//g;
 $perf ? print "$msg | $perf\n" : print "$msg\n";
 
 exit($stat);
+