Trigger dependences

[dzoleg]

I want to use trigger dependences in zabbix.
And I have many groups of hosts. One router exist in each group.
Other device's triggers is depend on router's triggers.
All triggers is based on "simple check" items.

The problem is when hosts is down or up the head (router's) triggers will not necessarily change a state
before slave triggers. Router's trigger can change state the second, third or sixth of hosts group...
for example:

2007.Oct.02 10:26:32 M135_ARM Alert for M135_ARM is in state ON High
2007.Oct.02 10:26:18 M135_Switch Alert for M135_Switch is in state ON High
2007.Oct.02 10:26:04 M135_Gateway Alert for M135_Gateway is in state ON High
2007.Oct.02 10:25:34 M135_WiFi Alert for M135_WiFi is in state ON High
2007.Oct.02 10:24:59 M135_Server Alert for M135_Server is in state ON High
2007.Oct.02 10:24:52 M135_Phone Alert for M135_Phone is in state ON High

or

2007.Oct.01 00:46:24 M147_Switch Alert for M147_Switch is in state OFF High
2007.Oct.01 00:46:21 M147_ARM Alert for M147_ARM is in state OFF High
2007.Oct.01 00:45:56 M147_Server Alert for M147_Server is in state OFF High
2007.Oct.01 00:45:30 M147_Gateway Alert for M147_Gateway is in state OFF High
2007.Oct.01 00:44:47 M147_Phone Alert for M147_Phone is in state OFF High

And I recieve not only one alert about head trigger "Mxxx_Gateway"

Number of hosts (monitored/not monitored/templates/deleted) 977(939/33/5/0)
Number of items (monitored/disabled/not supported)[trapper] 2235(2235/0/0)[0]
Number of triggers (enabled/disabled)[true/unknown/false] 2235(939/1296)[35/0/904]

[Alexei]

You may check status of the router much more frequently. It won't eliminate all messages about the hosts but should greatly decrease them.

[dzoleg]

I already do that. But that not decrease messages when hosts up. Because I can't change check order of unavailable hosts.

[dzoleg]

No solutions? Strange option in this case.

[Alexei]

I already do that. But that not decrease messages when hosts up. Because I can't change check order of unavailable hosts.

This is supposed to decrease number of messages when the hosts are up as well in ZABBIX 1.4.2 and higher.

[Siegfried]

Would it be possible for a future plan to do something like this:

- if trigger state change, check if dependency
- yes: test dependency
- no: state change

Something like this could ensure that dependencies would always be rechecked before actually enabling a check on the child services. I really don't know if it's something that could be feasible or hard to do, but it would be a nice feature At first I thought that's how it worked.

[Crazy Marty]

Would it be possible for a future plan to do something like this:

- if trigger state change, check if dependency
- yes: test dependency
- no: state change

Something like this could ensure that dependencies would always be rechecked before actually enabling a check on the child services. I really don't know if it's something that could be feasible or hard to do, but it would be a nice feature At first I thought that's how it worked.

Yes, this sounds like an ideal solution! Please implement this!

[Alexei]

Yes, this sounds like an ideal solution! Please implement this!

It was implemented some time ago, available in pre-1.4.3.