check_drbd

As other users have posted here, there is a bug in this script when use drbd version 8.4.

The output field identifier for role of node has changed in drbd version 8.3.
See: http://drbd.linbit.com/users-guide-8.3/ch-admin.html#id1369474

To use the script in version 8.3+, the role identifier needs to be changed from "st" to "ro".

Apply this patch:

--- check_drbd.orig 2014-12-29 18:03:37.000000000 +0100
+++ check_drbd 2014-12-29 18:03:59.000000000 +0100
@@ -217,7 +217,7 @@
if (/^s?(d+):.* cs:(w+)/) {
$cs{$1} = $2;
}
- if (/^s?(d+):.* st:(w+)//) {
+ if (/^s?(d+):.* ro:(w+)//) {
$st{$1} = $2;
}
if (/^s?(d+):.* ld:(w+)/) {

Can you fix it?

Regards
mirQ

The comparison of the local role and the expected role, with parameter "-r", doesn't work for me. It doesn't matter, which role were set with "-r", the result is always the same:

DRBD OK: Device 1 Connected UpToDate

The "--debug"-parameter shows maybe the bug. The expected role, set with "-r", were set in variable "DRBD Role", not in "DRBD Expect". The variable "DRBD Role" should contain the local Role and the Variable "DRBD Expect" the expected one, right?

So, is there a correction needed? Has someone another result?

--- /tmp/check_drbd.orig 2014-03-03 01:35:53.000000000 +0400
+++ check_drbd 2014-03-03 08:54:59.897998487 +0400
@@ -12,7 +12,7 @@
my $drbd_proc='/proc/drbd';
my $drbd_devices=0;
my ($drbd_expect, $drbd_role, $drbd_version, $debug_mode);
-my (%options, %cs, %st, %ld, %ds, %check, %warning, %critical);
+my (%options, %cs, %st, %ld, %ds, %oos, %check, %warning, %critical);

my $prog_name=basename($0);
my $prog_revision='0.5.3';
@@ -210,10 +210,17 @@
} else {
$input = "STDIN";
}
+ my $devnum;
while() {
if (/^version: (\d+).(\d+)/) {
$drbd_version = "$1.$2";
}
+ if (/^\s?(\d+):.*/) {
+ $devnum = $1;
+ }
+ if (defined($devnum) and /\soos:(\d+)/) {
+ $oos{$devnum} = $1;
+ }
if (/^\s?(\d+):.* cs:(\w+)/) {
$cs{$1} = $2;
}
@@ -248,11 +255,11 @@
#
my @devices;
if ($drbd_devices =~ /^all$/i) {
- for my $device ( keys %cs ) {
+ for my $device ( sort keys %cs ) {
push(@devices,$device);
}
} elsif ($drbd_devices =~ /^configured$/i) {
- for my $device ( keys %cs ) {
+ for my $device ( sort keys %cs ) {
next if ($cs{$device} eq "Unconfigured");
push(@devices,$device);
}
@@ -264,6 +271,9 @@
&myexit('UNKNOWN',"Could not find device $device");
}
$check{$device} = 1;
+ if (defined($oos{$device}) and $oos{$device} > 0) {
+ &myexit('WARNING',"Out of sync on device $device: $oos{$device} sectors");
+ }
}
if (int(keys %check) == 0) {
&myexit('UNKNOWN',"No configured devices found");

This plugin crucially misses the option "-d all" to check all the DRBD devices from the /proc/drbd file. Having to manually adapt which devices to check for each node/server is annoying.

I have four resources and following command is doing it's excellent job.

check_drbd -d 1,2,3,4 -r Primary -o Primary -w StandAlone -c WFConnection

When will be released checker for DRBD 8.4.x release?
Thanks, bye!
Morgan

Hello,

thanks for the nice script. I use it to monitor a primary/primary device in a XenServer Cluster.


Here my solution to a bug I had using drbd82-8.2.6-1.el5.centos Package from the CentOS Repository:

The state in /proc/drbd is defined as "ro:" and not as "st:" as the script expects. I added the following lines to subroutine "parse_proc" within the while loop:

if (/^s?(d+):.* ro:(w+)//) {
$st{$1} = $2;
}

Works perfect! Best regards,

pbr

Hello,

I really like this plugin. It is easy to use and the syntax is very well written. But I can not use it on DRBD 8.3.

The Plugin does not show the correct role. A short look and I realized that the plugin does not use the correct pattern. To detect if a DRBD volume is Primary or Secondary the "old" "st:" string is checked. but this string does not exists anymore.

Theifore I just changed in line 213 from:
if (/^s?(d+):.* st:(w+)//) {

to

if (/^s?(d+):.* ro:(w+)//) {

Now as far as I can see it works perfectly with version 0.8.3

Best Regards,
Phil