Dead Hosts Notifications

[khm]

In the settings you can activate "Dead hosts Notification" but you will get a message for every dead host.

Is there a chance to get a notification only for special hosts?

[TheWitness]

Feature requests go here: http://bugs.cacti.net

TheWitness

[amadill]

We just recently upgraded cacti to the latest version and added in the thold plugin. It is functional but we have a number of hosts that go up and down like yo-yos. It would be really useful to be able to enable/disable dead hosts detection. I've browsed through the code and I think I can hack it out on my own but I would like to follow acceptable practice when I do those mods.

The change that I'm wondering about is adding a field to the hosts table. Is that schema untouchable?

[cigamit]

We just recently upgraded cacti to the latest version and added in the thold plugin. It is functional but we have a number of hosts that go up and down like yo-yos. It would be really useful to be able to enable/disable dead hosts detection. I've browsed through the code and I think I can hack it out on my own but I would like to follow acceptable practice when I do those mods.

The change that I'm wondering about is adding a field to the hosts table. Is that schema untouchable?

So... you are wanting it per host? Because otherwise you can completely disable dead host alerting in the settings. You can also set the "Failure Count" in the settings (under Poller) to something a bit higher.

[amadill]

Yes, I need to disable it for most of the hosts that we monitor. (We are a WISP and I really don't need a text message every time a customer unplugs his CPE, but I do want one if one of our servers or AP's go down).
It would involve changes to host.php, global_settings.php, polling.php, and an additional field (thold_down?) in the host table. I'm just a bit reluctant to add an arbitrary field name to a well established schema.
The easiest place to add it is in the devices/change availability options.

[amadill]

As an experiment I added two fields to the availability options screen in host.php

diff host.php host.php.orig
404c404
< if (ereg("(availability_method|ping_method|ping_port|ping_timeout|ping_retries)", $field_name)) {
---
> if (ereg("(availability_method|ping_method|ping_port)", $field_name)) {

If the new field was in table host, a tiny change in polling.php where it does the downed host detection, and whatever support needed for the new field in the librarys. The code looks a bit more convoluted in the area where you add/edit a device in host.php.

My handicap is I don't do much coding in php and I'm not familiar with the code base.

[cigamit]

I will be adding an option for this in the next few days, possibly with an option to send to different email addresses per device. People have waited long enough I guess.

[amadill]

That would be awesome
BTW - browsing through the code gave an appreciation for how well all of this was put together. WD & thank you.

[TheWitness]

That would be awesome
BTW - browsing through the code gave an appreciation for how well all of this was put together. WD & thank you.

Compliments are welcomed, but the best form of sucking up is a cash donation...

TheWitness

[amadill]

<grin>In lieu of cash how about a donation of labor?</grin>
Updated Cacti 0.8.7g RPMs for Fedora Core 8 can be found at http://www.wiseoldcat.com/?q=node/9

[alexupton]

I seem to remember cacti/thold working like this in the past where if the monitor wasn't selected for a device, the thold plugin wouldn't send an e-mail... Am I wrong?

[amadill]

That might have been the case in a previous version. Another sysadmin claims that is how his works. I installed Monitor V1.2 and upgraded Thold to the latest version I could find, V0.4.3. and I still get Dead Host Notifications for systems that I excluded from monitor. But the monitor plugin does put a field in the host table that I can use. I added a couple of lines in poller.php that seems to be working.

Code: Select all

[includes]# diff polling.php polling.php.orig
177,183d176
< 
<                               if (api_plugin_is_enabled('monitor')) {
<                                       if ($host['monitor'] == '') {
<                                               continue;
<                                       }
<                               }
< 
301,306d293
<                       if (api_plugin_is_enabled('monitor')) {
<                               if ($host['monitor'] == '') {
<                                       continue;
<                               }
<                       }
< 

The flood of email from the unimportant systems has stopped.

[amadill]

I've been scratiching my head wondering why I am not getting messages when the host goes down. It turns out I am, it's the subject in the email that is wrong. At about line 327 in polling.php the routine where it detects hosts that have just gone down it has
$subject = read_config_option('thold_up_subject');
Shouldn't that be 'thold_down_subject' or something similar?