Yes, because it will probably cause confusing behaviour.
Adam.
Sent from BlueMail
Having the alert code say "Was device X online?" if false ->stop is bullshit magic?
Pretty much any other monitoring platform has the ability for parent/child relationships.
Spencer Ryan | Senior Systems Administrator | sryan@arbor.netArbor Networks+1.734.794.5033 (d) | +1.734.846.2053 (m)
On Fri, May 20, 2016 at 5:51 PM, Spencer Ryan <sryan@arbor.net> wrote:
You don't need to get fancy, if the device was offline when it was polled, treat it offline for that cycle. If it's up, it's up.Just like everything with the poller it would be a consistent 5 minutes.I think more of my colleagues here would trade a slight delay in alerting for not getting tens/hundreds of alerts when a site drops offline.
Spencer Ryan | Senior Systems Administrator | sryan@arbor.netArbor Networks+1.734.794.5033 (d) | +1.734.846.2053 (m)
On Fri, May 20, 2016 at 5:48 PM, Adam Armstrong <adama@memetic.org> wrote:
What, and delay alerts by a possible 5 minutes?
I know many if our users would hunt you down and stab you in the face for making such a suggestion.
And his do you know that the upstream device which was up at the start of the poller run is still up at the end?
Messing with poller order is a bad idea.
Adam.
Sent from BlueMail
On 20 May 2016, at 20:48, Spencer Ryan < sryan@arbor.net> wrote:Do the alerts get generated as each device polls? Just run the alerter when all pollers complete and take the parent device into account?
On May 20, 2016 3:44 PM, "Tom Laermans" < tom.laermans@powersource.cx> wrote:
.. or instilling intelligence into the poller to run polls based on "top level" (least amount of parent devices in tree) ... which is not impossible but definitely fiddly :-)
On 20/05/2016 21:42, Adam Armstrong wrote:
This is fiddly due to the way our poller runs as separated processes. You'd not be able to reliably suppress alerts this way, without also delaying them by up to 5 minutes.
Adam.
Sent from BlueMail
On 20 May 2016, at 19:07, Spencer Ryan < sryan@arbor.net> wrote:For example we have a testlab that has no UPS/Generator power that we monitor about 200 devices, when the building loses power we get those 200 alarms about devices being offline, when in reality there is a single 6500 as the gateway, and if that switch is offline, i know the rest will be.I've asked about this before but how hard would it be for you guys to implement alert dependencies?The easiest way I can think of doing it would be to simply let us set a "parent" device for anything, and if that device is offline don't send any alerts for downstream devices.
We don't want to disable the offline alarms because we do want to know when individual devices drop out.
Thanks!
Arbor Networks+1.734.794.5033 (d) | +1.734.846.2053 (m)
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list
observium@observium.org
http://postman.memetic.org/cgi-bin/mailman/listinfo/observium