topic Re: XMC Alarm on port congestion in ExtremeCloud IQ- Site Engine Management Center

XMC Alarm on port congestion

Anonymous — Wed, 16 Dec 2020 21:27:49 GMT

Hi,

Looking into configuring XMC to alarm on ports reporting congestion, possibly through the use of the following OID

extremePortCongDropPkts in the Extreme MIB with OID ".1.3.6.1.4.1.1916.1.4.14.1.1"

Couple of bits I need to figure out though is maybe configuring it on a threshold i.e. alarm if it exceeds a count greater than 100 and report switch and port, rather then for every drop - whether that's possible or not?

When looking at the trap configuration I don’t see an entry for that OID:

The idea is to give an early warning if congestion is happening and when it happens, rather then discovering it after the fact and or if the messages relate to possible reported issues.

Has anyone done anything like this before, and importantly know how to set it up?

Maybe there is a better way to report / alarm on it?

Many thanks in advance.

Re: XMC Alarm on port congestion

Dan15 — Tue, 05 Jan 2021 01:19:51 GMT

I assume the OID you have posted is not a trap rather an OID where you have to get the data actively with SNMP. Unfortunately XMC does not give you the option to do that.

Re: XMC Alarm on port congestion

Tomasz — Tue, 05 Jan 2021 20:15:59 GMT

Hi Martin,

Food for thoughts: would that be an option to achieve that through a syslog-based alarm? Threshold-based alarms for port utilization % is also doable if that does the job for you, remember to collect port statistics then (it may also be useful to restrict alarm relevance to a ‘device group’ which will in fact be a group of some ports you need to monitor).

Hope that helps,

Tomasz

Re: XMC Alarm on port congestion

Dan15 — Wed, 06 Jan 2021 16:48:01 GMT

Hi Tomasz,

The thing with threshold alarms is the statistics OIDs are not user configurable. Like in the example from Martin, if “extremePortCongDropPkts” is not already included in one of the preconfigured statistics, there’s no way to make XMC polling this OID and collect the data. For me this is a big lack in XMC compared to other tools.

Re: XMC Alarm on port congestion

StephanH — Wed, 06 Jan 2021 17:55:57 GMT

Hello,

what you can do as workaround is to write a short workflow calling the OID on a switch every few minutes. If the predefined value is exceeded, you can trigger an alarm via the workflow.

This is not really resource-saving but works fine. I you have only a few switches to monitor, it can be a option.

Re: XMC Alarm on port congestion

M_Nees — Thu, 07 Jan 2021 01:45:41 GMT

Hi,

i think workflow is a good way (bad for all non NMS-ADV Users).

But if you are using a workflow for this i think running CLI Queries for the whole job are more useful.

EXOS CLI:

“show port 1-48 congestion”

or maybe better

“sh port 1-48 congestion | exclude 0”

Re: XMC Alarm on port congestion

M_Nees — Thu, 07 Jan 2021 02:03:42 GMT

Better use this CLI Commands (otherwise CLI-Paging will only show first 12 ports):

show port 1-52 congestion no-refresh

or 

sh port 1-52 congestion no-refresh | exclude 0

Re: XMC Alarm on port congestion

M_Nees — Thu, 07 Jan 2021 02:09:16 GMT

Another way solving this Problem with XMC is:

define a custom FlexView for this OID (with legacy java tools).
running this Flexview for an automatic report (let’s say daily)

Disadvantage: Report have to be checked manually

Re: XMC Alarm on port congestion

M_Nees — Thu, 07 Jan 2021 03:07:31 GMT

Hi,

one last idea.

It was possible to run a UPM script periodically to check port congestion (CLI command). If there are some errors this UPM script will generate a log message.

Here is this example script which can be adapted:
https://extremeportal.force.com/ExtrArticleDetail?an=000080780

This log message can be forwarded directly to XMC or can generate a SNMP Trap which is send to XMC.

How to send an SNMP trap based on a log message
https://extremeportal.force.com/ExtrArticleDetail?an=000079593