CAPC - THRESHOLD - Current Established TCP CONNECTIONS

[Valeria Cunha]

Hello CA Team,

I would to know how can I make this threshold to send an alert when it have more than 86.000 connections (86 thousand connections) and clear equal or below to 72.000 connections.On CA Performance Center Server the 86.000 thousand connections would be 86.0?



Thank you,

Valéria

[Jeffrey Pinard]

If that 86k or 86 (sorry, are you using "." for US "," ?)

If 86 and 72, then the threshold profile rule is correct. 
If you want 86k and 72k, then use 86000 and 72000.

[Valeria Cunha]

Hello Jefrey,

I think that correct would be 86,000 in USA but when I try to put this value on CA Performance Threshold I got this error:

Because I want to have an alert with more than 82.000 connections (82 thousand).

If I put the 86000 like this would be 86.000 connections (86 thousand)?

thank you,

Valéria

[Jeffrey Pinard]

Don't use any "," or ".", just use raw 86000 or 72000.

[Valeria Cunha]

Hello Jeffrey,

Thank you for your reply. Unfortunately I saw other problem here.Meanwhile we had only one alarm and the correct would be 2 alarms.
On my environment we have 7 devices in a group,but only one had alarm,so Jeffrey I need to split the devices in many differents groups?
what is the difference between Window (sec):*and Duration (sec):*?

thank you,

Valéria

[Jeffrey Pinard]

You should need to only use 1 custom user group of all items you want to apply the threshold profile to.
Did you create a group of devices and choose "include children" checkbox?

Window is how long to look at data.  Duration is how long inside that window that the values need to satisfy the set or clear condition.

Example:  900 sec window and 600 sec duration means 2 out of the last 3 5-min data points must be > 86000 to create the event.  Or 2 of last 3 data point must be < 72000 to clear the event.
Using 300/300 means we only need 1 5-min data point to be > 86000 to create event and 1 data point < 72000 to clear event.

[Valeria Cunha]

Hello Jeffrey,

I have created the group and choose the option "include children" as you like said:


and you can see the 7 devices:


but the proxy 186 had more than 31000 connections and I didn't get the alarm here as you can see below the connections:



I have attached here the threshould:



thank you 

Valéria

[Jeffrey Pinard]

So since you are doing 1-min polling it seems, the 300 window/duration means you need 5 mins (aka 5 straight data points) of having the value >31000.   It also means you need 5 data points < 29000 before a clear is generated.

Is that view fixed to the group used for the thresholding?  or using the group context selected (aka All Groups)?
If All Groups, can you change the group context to the threshold group.  Do you see the lines for the expected group members?

[Valeria Cunha]

Hello Jeffrey,

My group is a fixed group and I can see the lines for the expected group members without problems,but now it seems that is working fine.

I can see the threshold alarm coming on ca spectrum and clearing in correct time. 

thank you for your supporting and explanation.

Valéria