1.4.1 Trigger expression NOT working !!

[BusteR81]

Hello, is there a bug with '&' expression or is there a bug in the .abschange(0) function ??

Trigger:
(Host02:cpmActiveDS0s.abschange(0)>10)&
(Host02:cpmActiveDS0s.last(0)<Host02:cpmActiveDS0s.prev(0))

Email Alert:
Sudden Drop in Total Active calls on Host02 on 2007.07.26 at 10:53:13 :
Current value = 6
Previous value = 7


Q: Why is the Trigger = TRUE and sent the above Email alert?

[BusteR81]

time function not available in 1.4.1 ?

tried using the useful .time(0) funtion but server told mi "time(0) not a valid function" so if such feature has not been added to 1.4.1, perhaps do an update on Zabbix Manual document in case anyone trys it and hit deadend.

Cheers
I do tink more in-depth tests should be done on triggers expressions especially complex ones with '&' and '|' etc (push trigger expression to limit) otherwise Zabbix is a Powerful tool.

[Alexei]

Trigger:
(Host02:cpmActiveDS0s.abschange(0)>10)&
(Host02:cpmActiveDS0s.last(0)<Host02:cpmActiveDS0s.prev(0))

Please could you post exact (with all {}) trigger expression? Thank you.

[BusteR81]

with {}

Trigger:
({Host02:cpmActiveDS0s.abschange(0)}>10)&
({Host02:cpmActiveDS0s.last(0)}<{Host02:cpmActiveD S0s.prev(0)})

[Alexei]

I cannot reproduce this with the latest pre 1.4.2. How the item cpmActiveDS0s is configured? Does it use delta?

[BusteR81]

juz a thought... no alarms

I cannot reproduce this with the latest pre 1.4.2. How the item cpmActiveDS0s is configured? Does it use delta?

cpmActiveDS0s returns total active calls from a snmp router (VoIP)

anyway i got it solved by working around and using simpler trigger expression, however i still have my doubts over complicated triggers expressions.

Eg...of how i feel over stability of trigger expression:

1+X = oK
(1+X)x2 = oK
((1+X)x2) + 3/X = still oK i guess
( ((1+X)x2) + (3/X) < 4 ) & ( ((5+2)x2) < ((6-2)x2) ) = ??? (can Zabbix do this maths equation with 100% confidence)

it is not that i dun trust Zabbix triggers but when maths get complicated, things might just go 'weird' Do u developers test with such complicated triggers expressions. oh well...

Cheers
just telling how i feel over 1.4.1 Trigger expression

[Alexei]

1+X = oK
(1+X)x2 = oK
((1+X)x2) + 3/X = still oK i guess
( ((1+X)x2) + (3/X) < 4 ) & ( ((5+2)x2) < ((6-2)x2) ) = ??? (can Zabbix do this maths equation with 100% confidence)

If you look at the server.c you could see that evaluation of trigger expressions is regression tested. Keep in mind operator precedence and you will be fine