Hello,
Has anyone been able to alert off of a value in a Universal Device Poller? I am running 9.1 SP1.
It's possible. What is it you are wanting to achieve?
I have a UDP that is polling an Acmepacket Session Border Controller for inbound and outbound concurrent sessions in separate tables and then I am adding them together with another poller that I call Total Concurrent Sessions. I want to alarm off of individual rows in the table if the values are greater than a certain value.
I am running into same issue with a few UnDP's I am trying to alert on as well (bgp state, f5 pool status, etc). Basically you can't pull out a single "row" of data that you want to alarm on. There is another thread on thwack for "alerting on bgp state". In there it appears TestTrack # 8047 is already in the works for getting this functionality. Not sure when this will be added to the product but it is extremely critical part of monitoring snmp tables.
If you poll the specific row, then you can alert on it. Would that work?
Hey Denny,That would work but the issue is like with any of the snmp table information, I would have to create individual pollers for every row of every table per device. The functionality we need here is we already easily get the entire table via UnDP get table, now we just need the functionality of alarming on a single row of data. If we try to include info in the alarm it will send an entire table column of info.
For instance lets take the standard bgp peer table. I made a nice UnDP table on all my nodes with bgp peers setup with info like IP of peer, AS number, status of peer, etc. If I try to write an alert when a bgp peer is down, if I tried to include the IP of the peer that is down, it sends all peers IP's for that node. The only thing I can do is say a very generic "A bgp peer is down....go check it out". That is fine but with the ability to pull out the one bgp peer that is down the alarm would be much more useful to our NOC with "172.16.x.x bgp peer is down on this router". With this I could have better alarms to send to the NOC so they dont have to take an extra step of finding out which "row" is in alarm state. Same premise woulc be applied to every other UnDP get table values I poll....F5 pools, Juniper Netscreen VPN Tunnels, and the list goes on and on.
I agree with Brian on this one. The tables that I am polling have around 60 rows. I was polling them with individual pollers for each row before we could poll for tables but the problem was that every time I added another realm to our Session Border Controller it would add another row and re-index all of the other rows so polling this as a table saves me an incredible amount of time.
Also, each row has a different limit that I would like to alarm off of so it also makes it difficult to alarm off of the entire table values as we have some rows I would alarm at 5 and some that I would alarm at 50.
Got it. We're already thinking about how we might implement this feature.
Almost 3 years on and this type of feature still isn't present. Not impressed
Andy.
Andy--
I've just let the PM know about this--he's researching this issue and will post back by Wednesday morning at the latest.
M
