[1.4.2] Zabbix sends message trigger=off when it didn't send trigger=on

[movenext]

Hi all,

I'm using Zabbix 1.4.2 and have set dependencies up (using the interface). It's all working properly; I have a router, switch and some servers. When the router goes down, all I get is the mail from zabbix that the trigger=on. Nothing is being send about the server being unreachable. This is the expected behaviour.

When the router comes back up, zabbix sends a message about the router being up (trigger=off) which is cool, but zabbix also sends messages about the triggers of the servers being off. But Zabbix never sent a message they were on (due to dependencies of course). How come it sends them when the trigger is on? I would expect just one message again about the router being available.

Is this supposed to work like this or am I missing something?

thanks,movenext

[Alexei]

Yes, it is supposed to work this way. Only if the router trigger is ON (down), ZABBIX won't send any messages about availability of servers behind the router.

When the router is back, ZABBIX starts sending new messages generated by the servers.

[michaeltje]

Yes, it is supposed to work this way. Only if the router trigger is ON (down), ZABBIX won't send any messages about availability of servers behind the router.

When the router is back, ZABBIX starts sending new messages generated by the servers.

isnt there a way to get rid of it? i have the same problem. I just want 1 message, just an example i have a server park of 100 servers. i monitor them with zabbix and the connection between the office and the server park drops. It will send me a nice message about the gateway being down (Perfect!) Then when the connection restores i will get 100 messages, i dont think thats the way how it should work?

Thanks!

[Alexei]

isnt there a way to get rid of it? i have the same problem. I just want 1 message, just an example i have a server park of 100 servers. i monitor them with zabbix and the connection between the office and the server park drops. It will send me a nice message about the gateway being down (Perfect!) Then when the connection restores i will get 100 messages, i dont think thats the way how it should work?

I agree, this has to be improved.

[michaeltje]

I agree, this has to be improved.

that would be great! When do u think u will improve this?

[Alexei]

that would be great! When do u think u will improve this?

It depends on many things, priorities, etc. It may even happen in 1.4.3, no promises.

[michaeltje]

It depends on many things, priorities, etc. It may even happen in 1.4.3, no promises.

I understand, its great enough that u will fix it! If u could keep us informed about this, because we really wait for this feature

Thanks for the great support!

[movenext]

Hi all,

Thanks for your responses!

Maybe you could add a check to the actions which only allows an action to be sent (trigger=off) when a previous action is sent for trigger=on. There already is a list of sent actions, so this should be possible.

Anyway, thanks in advance for improving Zabbix even more! Let me know if I can be of any help.

[Andreas Bollhalder]

Hello

I have a similar "problem" which I tried to get to work. I like to monitor our server and there services and net ports. To minimize messages sent, I made dependencies for a linux host like:

"SSH Port" depend on "SSH Service" depend on "ZABBIX Agent" depend on "ICMP Ping". With this, I get the following messages when

1. ICMP Ping down:
   ICMP Ping ON and OFF, ZABBIX Agent OFF
2. ZABBIX Agent down:
   ZABBIX Agent ON and OFF
3. SSH Service down:
   SSH Service ON and OFF
4. SSH Port down:
   SSH Port ON and OFF

I like to get rid for the message "ZABBIX Agent OFF" of list point 1.

I have implemented the items like

Description:		Key:		Update Interval
n_icmp-ping		icmpping	30
v_zabbix-agent-ping	agent.ping	30
s_ssh-process-count	proc.num[sshd]	60
n_tcp-ssh-reachable	ssh		60

and the triggers like

Name:					Expression:
host N: Ping is not returned		({host:icmpping.max(#3)}<1)|(({TRIGGER.VALUE}=1)&({host:icmpping.min(#14)}>0))

host V: Zabbix agent ping not received	({host:agent.ping.nodata(180)}=1)|(({TRIGGER.VALUE}=1)&({host:agent.ping.count(300)}<9))
Depends on 
host N: Ping is not returned

host S: OpenSSH is not running		({host:proc.num[sshd].max(#5)}<1)|(({TRIGGER.VALUE}=1)&({host:proc.num[sshd].min(#4)}>0))
Depends on 
host N: Ping is not returned
host V: Zabbix agent ping not received

host N: SSH is not reachable		({host:ssh.max(#7)}<1)|(({TRIGGER.VALUE}=1)&({host:ssh.min(#2)}>0))
Depends on
host N: Ping is not returned
host V: Zabbix agent ping not received
host S: OpenSSH is not running

So far, so well.

The problem I have is, that the second part of the trigger expression for the "ICMP Ping"

(({TRIGGER.VALUE}=1)&({host:icmpping.min(#14)}>0))

doesn't work. The triggers don't stay on TRUE for the expected time, it change immediatly to FALSE. Because the trigger for "ZABBIX Agent" is TRUE too, I get the OFF message when it changes to FALSE.

I would be happy if someone can point me back to the road.

Andreas

[remi]

Hi,

You could try change:

(({TRIGGER.VALUE}=1)&({host:icmpping.min(#14)}>0))

to:

(({TRIGGER.VALUE}=1)&({host:icmpping.max(#14)}>0))

This way, the trigger will go off if the last 14 values are all >0 instead of one of the last 14.

grtx,remi

[Andreas Bollhalder]

Hi,

You could try change:

(({TRIGGER.VALUE}=1)&({host:icmpping.min(#14)}>0))

to:

(({TRIGGER.VALUE}=1)&({host:icmpping.max(#14)}>0))

This way, the trigger will go off if the last 14 values are all >0 instead of one of the last 14.

grtx,remi

Hello remi

Thank a lot for the suggestion.

I made the test and changed the min to max. But then, the trigger stays forever on TRUE. I'm a little bit confused now.

As I read the docs:

max.(sec | #num) => Maximal value for period of time.

it should return THE highest value of the time period or the specified number of values. So the trigger would stay on TRUE until there are 14 values of "0". But then, the first part of the expression would evaluate to TRUE.

I don't know, why the min function doesn't work.

Now, there is smoke over my head.

Andreas

[Andreas Bollhalder]

Have changed the trigger to

({host:icmpping.max(#3)}=0)|(({TRIGGER.VALUE}=1)&({host:icmpping.min(#14)}=0))

and this seems to work fine. I definitly need some vacation right now.

Andreas

[michaeltje]

Hi,

You could try change:

(({TRIGGER.VALUE}=1)&({host:icmpping.min(#14)}>0))

to:

(({TRIGGER.VALUE}=1)&({host:icmpping.max(#14)}>0))

This way, the trigger will go off if the last 14 values are all >0 instead of one of the last 14.

grtx,remi

Try this ({host:icmpping.max(60)}<1)

[michaeltje]

It depends on many things, priorities, etc. It may even happen in 1.4.3, no promises.

Is there anything you can tell us about yet?
Thanks!

[Alexei]

Is there anything you can tell us about yet?
Thanks!

It is already in pre 1.4.3.