the action is not executed when the trigger condition is met

[seanxiao]

We are using Zabbix 4.2.8 to monitor our Huawei CE8860EI switch.
in zabbix we created two trigger with severity Disaster for this switch, one is for BGP state idle, the other one is for bgp state abnormal

expression:

{BJ-YZ-2FM1-IDC-D02:state.if[1.3.6.1.4.1.2011.5.25.177.1.1.2.1.5.0.1.1.1.4.10.2 7.0.1].last()}=1

for bgp state idle peering with 10.27.0.1, severity is disaster

expression :
problem: {BJ-YZ-2FM1-IDC-D02:state.if[1.3.6.1.4.1.2011.5.25.177.1.1.2.1.5.0.1.1.1.4.10.2 7.0.1].last()}<>6
for bgp state abnormal peering with 10.27.0.1, severity is disaster

expression:
recovery: {BJ-YZ-2FM1-IDC-D02:state.if[1.3.6.1.4.1.2011.5.25.177.1.1.2.1.5.0.1.1.1.4.10.2 7.0.1].last()}=6
for bgp state back to normal peering with 10.27.0.1

Action has been configured, it was supposed to be executed with the following condition:
1. a disaster is triggered
2. switch is in a specific group

Now we found that the action will be excuted when bgp state idle. SMS/phone ringing, etc.
but in the same switch, bgp state abnormal will not call action, it only display a problem in the problem dashboard. A weird thing is that the problem is triggered with desired severity Disaster, but it displays different severity other than Disaster in details page.
I've checked role/permission, action log (no action), alert media, etc. and I could not find any thing useful... Could you please help me out?

[tim.mooney]

Hi seanxiao-

My first guess was going to be that it was a permission problem: when the people you want to receive alerts don't have at least read access to the host group, the action won't run. That's a frequent source of actions not being executed. However, you've said you've checked permissions, and you've indicated that the other action for "bgp state idle" is executing, so it doesn't seem like it's a permission problem.

I'm not sure why it would display a different severity in the details page. There isn't some other template item/trigger that might be partially duplicating the behavior of your "disaster" alert, is there?

Unfortunately I can't read Chinese, so I'm only able to partially understand the graphics you've included. I'm not sure there would be anything that would be a "clue" that would help me even if I did read Chinese, though.

If you're comfortable reading C source code, you could try increasing the log level for some of the process types of zabbix_serverd (probably alerter and alert manager to start, though that may not be enough...) and then try track the resulting log messages through the code, to see if you can figure out why the trigger is firing but the action is not. It's really only an option if you're comfortable with the source code, though.

[seanxiao]

Hi seanxiao-

My first guess was going to be that it was a permission problem: when the people you want to receive alerts don't have at least read access to the host group, the action won't run. That's a frequent source of actions not being executed. However, you've said you've checked permissions, and you've indicated that the other action for "bgp state idle" is executing, so it doesn't seem like it's a permission problem.

I'm not sure why it would display a different severity in the details page. There isn't some other template item/trigger that might be partially duplicating the behavior of your "disaster" alert, is there?

Unfortunately I can't read Chinese, so I'm only able to partially understand the graphics you've included. I'm not sure there would be anything that would be a "clue" that would help me even if I did read Chinese, though.

If you're comfortable reading C source code, you could try increasing the log level for some of the process types of zabbix_serverd (probably alerter and alert manager to start, though that may not be enough...) and then try track the resulting log messages through the code, to see if you can figure out why the trigger is firing but the action is not. It's really only an option if you're comfortable with the source code, though.

dear friend,

thanks for your kind help.
yes there are multiple

I have no ability to read C source code, any log from the server that could be useful?

[tim.mooney]

I have no ability to read C source code, any log from the server that could be useful?

The debug-level log information that I was thinking of would be written to zabbix_server.log after you've enabled debug-level logging, but it's only really useful if you can understand C source code. That high level of log detail is primarily useful to track what's happening in the source code.