DX NetOps

Hi All,

We are getting some unusual bulk alerts for a specific time interval ,by anyway can we check is there is any packet loss from Spectrum servers to the end devices ? From spectrum server whether i need to check anything 

Regards, 

Hi Praveen,

We probably need more information regarding the Bulk Alerts/Alarms that you are seeing to better answer this question.

The term "Alert" is generally associated with traps in Spectrum. If this is trap related, you can see specific counters related to trap processing under the VNM model in OneClick. Go to the Component Detail pane - Information tab, and expand the "Trap Management" subview.  There you will find counters showing how many traps have been Received, Processed, Ignored, or Discarded due to Trap Storms.

If these are polled based alarms, and you are running a version of Spectrum other than 10.2.1, you could be experiencing a known issue where the SNMP Subsystem hangs. Are you seeing a lot of false "Management Agent Lost' alarms, or other alarms related to SNMP polling? If so, this is addressed in Spectrum 10.2.1. See TEC1776088 (https://www.ca.com/us/services-support/ca-support/ca-support-online/knowledge-base-articles.TEC1776088.html) for more information.

If you suspect a Network or NIC related problem, the best way to check to see if the server running Spectrum is dropping packets, would probably be a packet capture from the SpectroSERVER when the issue is happening.

Regards,

Brad

i am getting device down alerts for some set of servers after few secs alerts are getting cleared.

Are you on 10.2.1?  Are the alarms on Pingable model types? If so, there’s a PTF patch:

https://www.ca.com/us/services-support/ca-support/ca-support-online/knowledge-base-articles.tec1760694.html

If you need that, please open a case and we’ll get it for you.

Cheers!

Jay

Issue with spectrum 9.4 and with systemedge models.

Are the alarms clearing outside of the default 3 seconds?  If so, try increasing the DCM timeout to be greater than the amount of time it’s taking to clear the alarm.  For example, if the alarm is clearing in 16 seconds, set the DCM timeout to 20 seconds

Yes I agree  ,but it happens only one or two times ,by anyway i can trace ?

Sniffer trace is the easiest.  Just run the sniffer trace, load it up in wireshark, take a look at when you received the contact lost alarm.  Match up the request ids and take a look at the timestamps of when the packets were sent/received.

Cheers

Jay