not triggering on 100% loss ping

[micronemo]

I'm using Cacti 0.8.8a. Thold doesn't trigger devices when they go down.
i see a similar procedure for other cati version but don't work.
someone can help me ?
thanks very much

[tcprado]

I am having a similar problem. The graph has an open spot where no ping was received, thold plugin detects the device as down in the host status tab, the device tab says it is down but the threshold does not trigger. I set it for lower limit 1 and upper limit 5. Thold does not trigger when the device is unresponsive and ping times out.

what can I do? How should I set the threshold propperly?

[cigamit]

Thresholds will not trigger if the device is unresponsive, that is what down hosts messages are for. There are multiple reasons for this, but one of the biggest is that Cacti doesn't pass the plugin any data for failed pollings.

[tcprado]

Is there any way to work around this? or it is not intended by design?

[cigamit]

Thold should be sending down host alerts if the device is down (and you configured to receive them). The not alerts on any thresholds if the device is down, is by design and by limitation. Cacti doesn't pass us the data if the device is down, and we don't really want 15 thresholds (disk, etc...) going off if the device is down. A single device is down alert should be enough (which is why we have down host alerts).

[tcprado]

I see. I am monitoring network stability and was thinking of setting the ping latency thold so when a device doesn't respond, it logs the event and in the end of the month, I could count the number of treshold trespasses. but I understand your implementation. It makes much more sense.

[tcprado]

Where can I count the number of 'host down' events? I don't really care to be alerted, I just wante the area in the graph marked the way thold does. I achieved this by creating a specific graph template and using CDEF and the result is what I wanted. So I just want now to get the number of times the host was down, not caring for how long at the moment, as I can do that using my graph and it gives me the % available for the period selected.

[victorantunes]

I don't have access to any Cacti box right now, but the Downed Hosts area has a column that displays something like "Event Count", does it not?

If that's so, divide that number by your polling interval (in minutes, IIRC) and you'll have the total amount of times the poller (or is it Thold at this point? I've gotta check that source code) has found said device to be down, since the first moment it started being monitored by the Thold plugin.

However, it's important to note that the number keeps on increasing as long as the device is down. For example, the device stayed offline for 2 poller cycles, so the Event Count would receive a increment of +2, meaning you wouldn't be able to accurately get a "number of downtime occurrences" statistics kind of thing.

Hope it makes sense.

[tcprado]

I understand. I saw that column but it does not give me the number of times the host went down, it gives me the number of poller cycle it has been down since the last time it was down. What I would like is the number of times it went down on a given period. this tab does not take two dates as input to use as period.

But no worries, it already works well enough. I can overcome that by using the graph I created and counting the number of red areas. No worries.