Linux DNS Server monitoring (Bind,named)

[dima_dm]

Linux DNS Server (Bind,named) http://www.isc.org/software/bind

I use bind util
/usr/sbin/rndc stats
This command write statistics in file /var/named/chroot/var/named/named.stats (run from zabbix_bind_stats.pl)
Script /etc/zabbix/zabbix_bind_stats.pl write statistics for Zabbix in file /tmp/zabbix_named.stats.txt

Permissions on script /etc/zabbix/zabbix_bind_stats.pl

 chmod 4755 zabbix_bind_stats.pl
ls -l zabbix_bind_stats.pl
-rwsr-xr-x  1 root root 1585 Mar 18 12:06 zabbix_bind_stats.pl

Install perl-suidperl rpm!!!!!

Bind configuration:
/var/named/chroot/etc/named.conf

options {
	…
        statistics-file "/var/named/named.stats";
	…
	}

named.stats format is

+++ Statistics Dump +++ (1300812291)
success 1843046927
referral 96446157
nxrrset 188937348
nxdomain 658140493
recursion 1158184815
failure 1373105359
--- Statistics Dump --- (1300812291)

zabbix_named.stats.txt format is

date: 1301655089 Fri Apr  1 14:51:29 2011
success: 30547529
referral: 834977
nxrrset: 3616500
nxdomain: 14214245
recursion: 15199447
failure: 15493181

Zabbix_agent configuration:
/etc/zabbix/zabbix_agentd.conf

Timeout=30
## Named monitoring
UserParameter=named_stat,/etc/zabbix/zabbix_bind_stats.pl
UserParameter=named_success,egrep "success:" /tmp/zabbix_named.stats.txt| awk '{print $2}'
UserParameter=named_referral,egrep "referral:" /tmp/zabbix_named.stats.txt| awk '{print $2}'
UserParameter=named_nxrrset,egrep "nxrrset:" /tmp/zabbix_named.stats.txt| awk '{print $2}'
UserParameter=named_nxdomain,egrep "nxdomain:" /tmp/zabbix_named.stats.txt| awk '{print $2}'
UserParameter=named_recursion,egrep "recursion:" /tmp/zabbix_named.stats.txt| awk '{print $2}'
UserParameter=named_failure,egrep "failure:" /tmp/zabbix_named.stats.txt| awk '{print $2}'
UserParameter=named_udp,netstat -nua | grep :53 | wc -l
UserParameter=named_tcp,netstat -nta | grep :53 | wc -l
## End Named monitoring

Example:

[frater]

This is a great idea!
But if local scripts are involved already I prefer to let them handle all itself without the agent and use zabbix_sender.

IMHO it is better, but maybe someone can explain me otherwise.
No change in zabbix_agentd.conf

# ln -s /usr/local/sbin/zabbix-bind /etc/cron.min/zabbix-bind
# chmod +x /usr/local/sbin/zabbix-bind
# cat /usr/local/sbin/zabbix-bind

#!/bin/sh
export PATH=/usr/local/sbin:/usr/local/bin:/sbin:/bin:/usr/sbin:/usr/bin:/root/bin

ZAB_CONF=/etc/zabbix/zabbix_agentd.conf
ZAB_BIN=/usr/local/sbin/zabbix_sender
RNDC_BIN='/usr/sbin/rndc'
STATS='/var/cache/bind/named.stats'
FTMP1=`mktemp`

if $RNDC_BIN stats ; then
  if [ -s ${STATS} ] ; then
    tail -n12 ${STATS} 2>/dev/null | grep -A12 '^+++' | grep '^[a-z]' | sed 's/.*/- named_&/' >${FTMP1}
    if tty >/dev/null ; then
      $ZAB_BIN -vv -c ${ZAB_CONF} -i ${FTMP1}
    else
      $ZAB_BIN -c ${ZAB_CONF} -i ${FTMP1} >/dev/null
    fi
  elif tty >/dev/null ; then
    echo "${STATS} does not exist!" >&2
  fi
fi
rm -f ${FTMP1}

I do hope the OP welcomes this different approach as I have no intention to offend him, nor to hijack his thread.
It does need some changes in the template as the items are now of type 'zabbix_trapper' and not 'zabbix_agent'.

[dima_dm]

IMHO it is better, but maybe someone can explain me otherwise.

In such way I can supervise as the script is executed. If there will be any problems, I at once shall receive a error code on Zabbix (wrong permissions, absence of necessary programs and modules etc).

[frater]

In such way I can supervise as the script is executed. If there will be any problems, I at once shall receive a error code on Zabbix (wrong permissions, absence of necessary programs and modules etc).

{Template_App_named:named_success.nodata(180)}=1

I added a trigger that will react when it has less than 10 resolves per second during work hours and less than 1 off hours (both disabled by default).

Another trigger that will notify you if it didn't receive data the last 3 minutes. The latter will also notify you the script isn't working as it should.

[dima_dm]

{Template_App_named:named_success.nodata(180)}=1

I know and use this trigger too.
But I can send error code also with detail problem description.
See example:

http://www.zabbix.com/forum/showthread.php?t=15691

0 OK
1 VMWare client error
2 Status file write error
3 bad username or password
4 Can't connect to ESXi
5 Can't get data from ESXi

[frater]

I know and use this trigger too.
But I can send error code also with detail problem description.
See example:

http://www.zabbix.com/forum/showthread.php?t=15691

0 OK
1 VMWare client error
2 Status file write error
3 bad username or password
4 Can't connect to ESXi
5 Can't get data from ESXi

I may not have a good idea how you work with zabbix, but I can't see the use of it. It works or it doesn't and if it doesn't there's something wrong with the config and you need to troubleshoot it. That machine is being monitored by an agent and if there's something wrong with the machine I will get other messages.
I'm getting messages if /etc/zabbix/zabbix_agentd.conf changes, there are errors in /var/log/zabbix/zabbix_agentd.conf.

Anyway... Thanks for the idea. I guess we have all different ideas about what's a good way to implement things.
When I saw an example of someone using the zabbix_sender I had some difficulties to implement them for the first time. I now know the pitfalls and prefer it in some situations.

[Zophren]

Hello,

Thanks you for the script but i can't make it work.

Indeed this the result of "more /tmp/zabbix_named.stats.txt" after executed the perl script:

date: 1300801289 Tue Mar 22 14:41:29 2011
success:
referral:
nxrrset:
nxdomain:
recursion:
failure:

I think my named.states format is different of your .

++ Name Server Statistics ++
12278 IPv4 requests received
126 requests with EDNS(0) received
12278 responses sent
126 responses with EDNS(0) sent
4224 queries resulted in successful answer
12273 queries resulted in authoritative answer
5 queries resulted in non authoritative answer
4049 queries resulted in nxrrset
4005 queries resulted in NXDOMAIN
4 queries caused recursion

The regex are differents ...

I can change the regex, but i don't understand the role of $position variable in your perl script.

Can you help me ?


Many thanks !

[dima_dm]

Yes. My named.stats format is

+++ Statistics Dump +++ (1300812291)
success 1843046927
referral 96446157
nxrrset 188937348
nxdomain 658140493
recursion 1158184815
failure 1373105359
--- Statistics Dump --- (1300812291)

but i don't understand the role of $position variable in your perl script.

$position is used to clear file named.stats if size of file named.stats >$maxsize.
And $position is used for performance optimization. We will use only last statistics sample. For what reason will we parse all file named.stats, when we can process only last statistics sample (I am assume, that one statistics sample can't take more than $data=300 Byte)?

[Shriram.Fowkar]

Linux DNS Server monitoring (Bind,named)

Hi,

Thanks for the perl script. As suggested in the script I upgraded my perl-suidperl to version 5.8.5-36.

However when I run the script manually I get proper output in /tmp/zabbix_named.stat.txt. Same as you get in yours.

But when the same script actually gets executed through the agent it throws blank values. I am puzzled.

Any ideas why it must be happening ?

Thanks.

--
Shriram

[dima_dm]

run as
su - zabbix
/etc/zabbix/zabbix_bind_stats.pl

[Shriram.Fowkar]

Linux DNS Server monitoring(Bind,named)

Thanks for the reply.

I have disabled my zabbix userid shell with /sbin/nologin.

I changed that to a valid shell. Zabbix user is without password. I have set the password. My /etc/rndc.conf was having read permission to group named, so added zabbix user to named group in /etc/group. After this when I do

su - zabbix
/opt/zabbix/sbin/zabbix_bind_stats.pl

It worked and gave me proper output in /tmp/zabbix_named_stats.txt

However when the same thing runs through agent still gives me blank values.

I reverted back again the zabbix user to /sbin/nologin shell and removed zabbix password.

Still the same status. In the zabbix_agent.log it throws the error :

rnd: error: /etc/rndc.conf:33: open: /etc/rndc.key: permission denied.

I even tried changing the permissions of /etc/rndc.conf to 644, but still the same status.

[Shriram.Fowkar]

Linux DNS Server monitoring(Bind,named)

I restarted the zabbix_agent on the client and now the data is coming correct.

Thanks for all the help and this beautiful script with template.

[Superkikim]

named.stats new format

Hey. Tx for this script.

Since you made it, the format of named.stats changed.

Could anyone adapt the script to the new format ? I'm not that good in Perl scripting

Would appreciate any help.

Here is the new format:

+++ Statistics Dump +++ (1359329765)
++ Incoming Requests ++
                  20 QUERY
++ Incoming Queries ++
                   9 A
                   3 PTR
                   8 AAAA
++ Outgoing Queries ++
[View: default]
[View: _bind]
++ Name Server Statistics ++
                  20 IPv4 requests received
                   9 recursive queries rejected
                  20 responses sent
                   7 queries resulted in successful answer
                  11 queries resulted in authoritative answer
                   4 queries resulted in nxrrset
                   9 other query failures
++ Zone Maintenance Statistics ++
                  18 IPv4 notifies sent
++ Resolver Statistics ++
[Common]
[View: default]
[View: _bind]
++ Cache DB RRsets ++
[View: default]
[View: _bind (Cache: _bind)]
++ Socket I/O Statistics ++
                   5 UDP/IPv4 sockets opened
                   2 TCP/IPv4 sockets opened
                   1 TCP/IPv6 sockets opened
                   4 UDP/IPv4 sockets closed
                   2 TCP/IPv4 sockets closed
                   3 TCP/IPv4 connections accepted
                  12 UDP/IPv4 send errors
++ Per Zone Query Statistics ++
--- Statistics Dump --- (1359329765)

[skygge]

I agree, stats file has different format, script doesn't work.
Please update script... thanx!

My stats file is in the attachment.

[dima_dm]

I agree, stats file has different format, script doesn't work.
Please update script... thanx!

My stats file is in the attachment.

See Linux DNS Server monitoring LLD (Bind 9.6.X and high,named)

https://www.zabbix.com/forum/showthread.php?t=38970