Is it possible to have Zabbix monitor ntp?

[jackiebaron]

OK, I'm quite the newbie and this question is probably a little vague or has probably been answered but I do need help.

I have some ntp servers and each day I have to check the health of the ntp on them. Not that the ntp daemon is running..as simply ps -ef could do that. But is there actually any kind of Zabbix module that checks the health of the ntp service?
When I execute ntpq -p I should get something like the following:

$ ntpq -p
remote refid st t when poll reach delay offset jitter
================================================== ============================
192.168.140.21 .MRS. 1 u 1 16 377 0.970 0.057 0.275
oPPS(0) .PPS. 0 l 3 16 377 0.000 -0.023 0.000
+SHM(0) .shm0. 0 l 9 16 377 0.000 -0.015 0.000



That tells me that it's ok when I have those 3 entries.

I'm still just getting up to speed on Zabbix so please don't bombard me with serious stuff that only a Zabbix master would understand.

Thanks

[BDiE8VNy]

You can extend the Zabbix agent (as well as proxy and server) by custom functions via User parameters or loadable modules.
See User parameter one-liners to get some ideas regarding User parameters.

[harmonica]

OK, I'm quite the newbie and this question is probably a little vague or has probably been answered but I do need help.

I have some ntp servers and each day I have to check the health of the ntp on them. Not that the ntp daemon is running..as simply ps -ef could do that. But is there actually any kind of Zabbix module that checks the health of the ntp service?
When I execute ntpq -p I should get something like the following:

$ ntpq -p
remote refid st t when poll reach delay offset jitter
================================================== ============================
192.168.140.21 .MRS. 1 u 1 16 377 0.970 0.057 0.275
oPPS(0) .PPS. 0 l 3 16 377 0.000 -0.023 0.000
+SHM(0) .shm0. 0 l 9 16 377 0.000 -0.015 0.000



That tells me that it's ok when I have those 3 entries.

I'm still just getting up to speed on Zabbix so please don't bombard me with serious stuff that only a Zabbix master would understand.

Thanks

Hi,

If you want to check if the service is runing use proc.num[ntpd]

If you want to check the service state use ntpstat | echo $?
0 ⇒ synchronised
1 ⇒ not synchronised
2 ⇒ indeterminant

[jackiebaron]

You mean do a

zabbix_get -s <ntp machine> -p10050 -k proc.num[ntpd] ??

[harmonica]

You mean do a

zabbix_get -s <ntp machine> -p10050 -k proc.num[ntpd] ??

Hi.

Create item proc.num[ntpd] in every NTP server.

[jackiebaron]

Hi.

Create item proc.num[ntpd] in every NTP server.

OK harmonica, I really appreciate your help but like my original post stated I am a newbie so don't assume I even know what an "item" is.

You mentioned if I want to check if ntp is running then I just execute this from the Zabbix server:

zabbix_get -s <node> -p10050 -k proc.num[ntpd]

and that will tell me how many ntpd processes are running. That's fine. I can do that on the node with a simple ps -ef |grep ntp as well. But this is manual.

I need for the agent to monitor the ntp service by itself and have a message pop up on the Zabbix management server when the service is DOWN.

How come when I kill http on the managed node I get a message on the Zabbix server saying "http is down", but when I kill ntp I get nothing?

[BDiE8VNy]

[...] don't assume I even know what an "item" is.

I strongly encourage to read the online documentation first. Without knowing the naming convention and understanding the underlying concepts, it's likely to become frustrated quickly.

As a starting point you might take a look at these chapters (incl. sub chapters):

https://www.zabbix.com/documentation/2.4/manual/concepts

https://www.zabbix.com/documentation/2.4/manual/quickstart

https://www.zabbix.com/documentation/2.4/manual/config

[cferraro]

Hi,

If you want to check if the service is runing use proc.num[ntpd]

If you want to check the service state use ntpstat | echo $?
0 ⇒ synchronised
1 ⇒ not synchronised
2 ⇒ indeterminant

I replaced the pipe with a semicolon above. When using a pipe I always got an exit of 0 regardless of the service state. The semicolon strings the 2 commands for expected results.