mirrors
/
sanoid
mirror of https://github.com/jimsalterjrs/sanoid.git
2
0
Fork 0
Code Issues Projects Releases Wiki Activity

allow monitor-health to optionally check zpool capacity too by providing limits along the flag

This commit is contained in:
Christoph Klaffl 2018-02-26 18:16:06 +01:00
parent 1f64c9c35a
commit 6c695f1a86
No known key found for this signature in database
GPG Key ID: FC1C525C2A47CC28
2 changed files with 66 additions and 25 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
README.md
View File

@ -60,7 +60,7 @@ Which would be enough to tell sanoid to take and keep 36 hourly snapshots, 30 da
+ --monitor-health
This option is designed to be run by a Nagios monitoring system. It reports on the health of the zpool your filesystems are on. It only monitors filesystems that are configured in the sanoid.conf file.
This option is designed to be run by a Nagios monitoring system. It reports on the health of the zpool your filesystems are on. It only monitors filesystems that are configured in the sanoid.conf file. Optionally it can check capacity limits too by appending them like '=80,95" (>= 80% warning, >=95% critical).
+ --force-update

89
sanoid
View File

@ -17,7 +17,7 @@ use Time::Local; # to parse dates in reverse
my %args = ("configdir" => "/etc/sanoid");
GetOptions(\%args, "verbose", "debug", "cron", "readonly", "quiet",
"monitor-health", "force-update", "configdir=s",
"monitor-health:s", "force-update", "configdir=s",
"monitor-snapshots", "take-snapshots", "prune-snapshots"
) or pod2usage(2);
@ -51,7 +51,7 @@ my @params = ( \%config, \%snaps, \%snapsbytype, \%snapsbypath );
if ($args{'debug'}) { $args{'verbose'}=1; blabber (@params); }
if ($args{'monitor-snapshots'}) { monitor_snapshots(@params); }
if ($args{'monitor-health'}) { monitor_health(@params); }
if (defined($args{'monitor-health'})) { monitor_health(@params); }
if ($args{'force-update'}) { my $snaps = getsnaps( \%config, $cacheTTL, 1 ); }
if ($args{'cron'}) {
@ -76,13 +76,32 @@ sub monitor_health {
my @messages;
my $errlevel=0;
my %capacitylimits;
# if provided, parse capacity limits
if ($args{'monitor-health'} ne "") {
my @values = split(',', $args{'monitor-health'});
if (!check_capacity_limit($values[0])) {
die "ERROR: invalid zpool capacity warning limit!\n";
}
$capacitylimits{"warn"} = $values[0];
if (scalar @values > 1) {
if (!check_capacity_limit($values[1])) {
die "ERROR: invalid zpool capacity critical limit!\n";
}
$capacitylimits{"crit"} = $values[1];
}
}
foreach my $path (keys %{ $snapsbypath}) {
my @pool = split ('/',$path);
$pools{$pool[0]}=1;
}
foreach my $pool (keys %pools) {
my ($exitcode, $msg) = check_zpool($pool,2);
my ($exitcode, $msg) = check_zpool($pool,2,\%capacitylimits);
if ($exitcode > $errlevel) { $errlevel = $exitcode; }
chomp $msg;
push (@messages, $msg);
@ -748,6 +767,8 @@ sub check_zpool() {
my $pool=shift;
my $verbose=shift;
my $capacitylimitsref=shift;
my %capacitylimits=%$capacitylimitsref;
my $size="";
my $used="";
@ -799,15 +820,21 @@ sub check_zpool() {
if ($health eq "ONLINE" ) {
$state = "OK";
# check capacity
my $capn = $cap;
$capn =~ s/\D//g;
if (%capacitylimits) {
# check capacity
my $capn = $cap;
$capn =~ s/\D//g;
if ($capn >= 80) {
$state = "WARNING";
}
if ($capn >= 95) {
$state = "CRITICAL";
if ($capacitylimits{"warn"}) {
if ($capn >= $capacitylimits{"warn"}) {
$state = "WARNING";
}
}
if ($capacitylimits{"crit"}) {
if ($capn >= $capacitylimits{"crit"}) {
$state = "CRITICAL";
}
}
}
} else {
if ($health eq "DEGRADED") {
@ -911,6 +938,20 @@ sub check_zpool() {
return ($ERRORS{$state},$msg);
} # end check_zpool()
sub check_capacity_limit() {
my $value = shift;
if ($value !~ /^\d+\z/) {
return undef;
}
if ($value < 1 || $value > 100) {
return undef;
}
return 1
}
######################################################################################################
######################################################################################################
######################################################################################################
@ -1081,19 +1122,19 @@ Assumes --cron --verbose if no other arguments (other than configdir) are specif
Options:
--configdir=DIR Specify a directory to find config file sanoid.conf
--configdir=DIR Specify a directory to find config file sanoid.conf
--cron Creates snapshots and purges expired snapshots
--verbose Prints out additional information during a sanoid run
--readonly Simulates creation/deletion of snapshots
--quiet Suppresses non-error output
--force-update Clears out sanoid's zfs snapshot cache
--cron Creates snapshots and purges expired snapshots
--verbose Prints out additional information during a sanoid run
--readonly Simulates creation/deletion of snapshots
--quiet Suppresses non-error output
--force-update Clears out sanoid's zfs snapshot cache
--monitor-health Reports on zpool "health", in a Nagios compatible format
--monitor-snapshots Reports on snapshot "health", in a Nagios compatible format
--take-snapshots Creates snapshots as specified in sanoid.conf
--prune-snapshots Purges expired snapshots as specified in sanoid.conf
--monitor-health[=wlimit[,climit]] Reports on zpool "health", in a Nagios compatible format
--monitor-snapshots Reports on snapshot "health", in a Nagios compatible format
--take-snapshots Creates snapshots as specified in sanoid.conf
--prune-snapshots Purges expired snapshots as specified in sanoid.conf
--help Prints this helptext
--version Prints the version number
--debug Prints out a lot of additional information during a sanoid run
--help Prints this helptext
--version Prints the version number
--debug Prints out a lot of additional information during a sanoid run