Zabbix 1.1.3 action {TRIGGER.NAME}: {STATUS} wrong after status change

[netod]

Hi guys, I'm using version 1.1.3 for both server and agentd. I have an action with the following template:

Trigger severity = "High"
Trigger severity = "Disaster"
Trigger value = "ON"


Severity: {TRIGGER.SEVERITY}
Time: {TIME} - {DATE}
Host: {HOSTNAME}
IP: {IPADDRESS}

{TRIGGER.NAME}: {STATUS}
Trigger key: {TRIGGER.KEY}
Value: {{HOSTNAME}:{TRIGGER.KEY}.last(0)}

when the trigger changes to ON, evertything is fine and the email reports:

Severity: High
Time: 12:07:20 - 2006.11.14
Host: XXXXXXXXXXXXXXXXXXXX
IP: xxx.xxx.xxx.xxx

Server XXXXXXXXXXXXXXXXXX is unreachable: ON
Trigger key: status
Value: 2

when the trigger changes back to OFF,

I have another alert template with the following:

Trigger severity = "High"
Trigger severity = "Disaster"
Trigger value = "OFF"

Severity: {TRIGGER.SEVERITY}
Time: {TIME} - {DATE}
Host: {HOSTNAME}
IP: {IPADDRESS}

{TRIGGER.NAME}: {STATUS}
Trigger key: {TRIGGER.KEY}
Value: {{HOSTNAME}:{TRIGGER.KEY}.last(0)}



When the status changes to OFF I get an email with the following:


Severity: High
Time: 12:08:52 - 2006.11.14
Host: XXXXXXXXXXXXXXXXXXXXXX
IP: xxx.xxx.xxx.xxx

Server XXXXXXXXXXXXXXXXXXX is unreachable: ON <------------------- Should be OFF?
Trigger key: status
Value: 0 <------------ notice the value which indicates that the trigger is OFF...


Has anyone else seen this kindda problem? Seems like an issue with states, but I can't work out why its happening. Any help would be appreciated.

[Alexei]

I cannot confirm this, everything works fine here in our test environment.

[disgruntleddutch]

From what I can tell, 0 is what you are supposed to be getting back for a status value if its seen as available, whereas 2 means its not available. I get 0 back for status on all my monitored hosts.

[netod]

From what I can tell, 0 is what you are supposed to be getting back for a status value if its seen as available, whereas 2 means its not available. I get 0 back for status on all my monitored hosts.

yeah, the action email seems to contain the correct value of the trigger (if 0, host is OK, if 2 unreachable) But the actual status is not correct, status should be set to OFF...I get ON in both emails. Can anyone else confirm this kind of behavior?

[netod]

I cannot confirm this, everything works fine here in our test environment.

I tested this again with a different host running version 1.1.3 agentd. The ON trigger specified above sends out 3 repeats. The OFF trigger acts only once.


Content of ON email

Severity: High
Time: 14:59:36 - 2006.11.15
Host: XXXXXXXXXXXXXXXXXXX
IP: 202.xxx.xxx.xxx

Server XXXXXXXXXXXXXXXXXXXX is unreachable: ON <----------- CORRECT value
Trigger key: status
Value: 2 <------------------ CORRECT VALUE


Content of OFF email:

Severity: High
Time: 15:31:41 - 2006.11.15
Host: XXXXXXXXXXXXXXXXXXX
IP: 202.xxx.xxx.xxx

Server XXXXXXXXXXXXXXXXXXX is unreachable: ON <------------ INCORRECT Value
Trigger key: status
Value: 0 <-------------------- CORRECT value


These tests are executed by manually stopping and strating the zabbix agentd on the given host. Can you think of any reason why this would be happening?

[Alexei]

Please could you post your trigger expression as well?

[netod]

Please could you post your trigger expression as well?

Sure.. The trigger in question is for a host unreachable scenario

({Unix_t:status.last(0)}=2)&({Unix_t:status.min(60 0)}=2)&({Unix_t:system.uptime.nodata(600)}=1)
( Above is redundant)

[Alexei]

Sure.. The trigger in question is for a host unreachable scenario

({Unix_t:status.last(0)}=2)&({Unix_t:status.min(60 0)}=2)&({Unix_t:system.uptime.nodata(600)}=1)
( Above is redundant)

Your trigger depends on nodata() as well, it means that it may become OFF regardless of value of the status. I wouldn't the status with function nodata(), use a processor load item, TCP ping, whatever. It makes no sense using the status with 'nodata'.

[netod]

Your trigger depends on nodata() as well, it means that it may become OFF regardless of value of the status. I wouldn't the status with function nodata(), use a processor load item, TCP ping, whatever. It makes no sense using the status with 'nodata'.

Well I thought the way the trigger worked in words is this:

if ( last status = 2 ) AND (minimum value of status in last 600 seconds = 2) AND ( uptime value has not received data in the last 600 seconds = 1 ) then call the action.

This trigger would become true when the host status is 2 AND host status is 2 in last 600 seconds AND there was no uptime data received from host in last 600 seconds.

Am I wrong on this?

The nodata function is on the item uptime, this should make sense since uptime is polled and not an internal value. No?

[Alexei]

My apologies! I didn't notice the 'uptime', I thought 'nodata' was defined for the 'status'. You're absolutely right!

[netod]

My apologies! I didn't notice the 'uptime', I thought 'nodata' was defined for the 'status'. You're absolutely right!

So that means that the trigger should not be the main cause of this weirdness? Not sure how else I can try and debug it short of going through the code. What I might do is recompile 1.1.4 and see if that helps with this issue.

[Alexei]

Yes, something wrong is here. More testing will be performed prior to release of 1.1.5. I will keep you informed in this thread.

[netod]

Yes, something wrong is here. More testing will be performed prior to release of 1.1.5. I will keep you informed in this thread.

I have just upgraded to version 1.1.4 (server only) and I'm having the same problem. When the trigger becomes ON:

Severity: High
Time: 10:01:27 - 2006.11.17
Host: dev.XXXXXXXXXX.com.au
IP: 203.xxx.xxx.xxx

Server dev.XXXXXXXXXX.com.au unreachable: ON <--------CORRECT
Trigger key: status
Value: 2 <---------- CORRECT


Severity: High
Time: 10:02:45 - 2006.11.17
Host: dev.XXXXXXXXXX.com.au
IP: 203.xxx.xxx.xxx

Server dev.XXXXXXXXXX.com.au is unreachable: ON <-------- should be OFF
Trigger key: status
Value: 0 <--------- CORRECT

[bbrendon]

Is this problem discussed here related to repeats? I'm on 1.1.2 and if I use repeats setup when using an action with Trigger Value = ON, I'll get repeats even if the the Trigger Value = OFF before the repeat would run.

[peter_field]

I third this

I am intermittenly experiencing this also. I haven't looked into it too far, but I cant see why it works properly sometimes and not others. Perhaps something to do with the trigger going into UNKNOWN? Just throwing it out there?

Services Not Running: ON <-- Correct
2006.11.27 23:50:26
Trigger key: uServicesNotRunning
Value: AE Tax Publisher <-- Correct
Host: ffsrv

Services Not Running: ON <-- Should be OFF.
2006.11.28 11:50:26
Trigger key: uServicesNotRunning
Value: None <-- Correct
Host: ffsrv

<edit to post more detail>
Sorry, here are the details of my config.

The action is generic and works fine for most triggers. The trigger that has the most trouble (it does work properly sometimes) is listed below, one thing I have noticed with this trigger is it's status is quite often UNKNOWN (which is another matter for another thread).

Item Configuration:
Desc: Products Outdated
Type: Zabbix Agent
Key: uProdChk
Type: INT64
Units: (none)
Mult: Do not use
Interval/Hist/Trends: 86400/30/365
Status: Monitored
Store: As is
Show: As is

Trigger configuration:
Name: Products Outdated
Expr: {Windows_t:uProdChk.last(0)}>0
Deps: no dependences
Severity: Average
No comments/URL/not disabled

Action configutation:
Type: send message
Source: trigger
Conditions: Trigger severity >= "Warning"
Send to: single user
User: admin
Subject: FF:{STATUS}:{HOSTNAME}:{TRIGGER.NAME}
Message:
{TRIGGER.NAME}: {STATUS}
{DATE} {TIME}
Trigger key: {TRIGGER.KEY}
Value: {{HOSTNAME}:{TRIGGER.KEY}.last(0)}
Host: {HOSTNAME}

ON = Problem is occurring.
OFF = Problem has been rectified.
Repeat: No repeats
Status: Enabled

I hope this helps.

Peter