packet loss alert

[NeilR]

Anyone have any insigh tinto what the of "performance task" "Interface In/Outbound Packet Loss Rate"  is measuring? I seem to be getting a number of these Performance Multilevel Alerts on some interfaces, but when I check at the switch there are no drops or other errors logged on the interface

OID 1.3.6.1.4.1.25506.4.1.2.2.0.8

It's a nuisance more than anything, but still curious

[jguse]

Hello Neil,

The Interface In/Outbound Packet Loss Rate is referring to the ifInDiscards/ifOutDiscards:

1.3.6.1.2.1.2.2.1.13.x and 1.3.6.1.2.1.2.2.1.19.x

I think you are familiar with discards - this is a sort-of hidden counter on the switch, and seeing some discards on an interface is nothing abnormal. Switches will discard some packets (like drops due to ACL filtering or many other reasons) during normal operation.

[pattap]

Hi @jguse @NeilR , i know it's a bit old but it might be useful for others.

I'm getting the same traps and it's a lot of them! I think it started happening since I setup a port group on IMC which I then removed but I stil receive the traps. threshold value seems ridiculous. Is there a way of disabling them on switches, comware 7 let's say. Can't see much in snmp trap enable config part that would relate to this.

The instance "[Interface:GigabitEthernet1/0/2]" of "performance task" "Interface Inbound Packet Loss Rate" is in threshold 2 range (">=0.01%"), current value is "0.591%".

[jguse]

Hello,

These are not SNMP Traps so you cannot disable them on the switch side. iMC Performance Management is polling the devices on the OIDs I mentioned above when you configure the "Interface Inbound/Outbound Packet Loss Rate" monitoring on your devices. I would not recommend using these performance monitors for the reasons listed previously... a better monitor to use would be the inbound/outbound error rate, as that triggering theshold alarms could indicate an actual issue on your switch/interfaces.

That said, you can also adjust the thresholds for this packet loss rate monitor individually or globally.

Globally - go to Performance Management > Global Index Settings, find the respective index (it's under Other Interface Statistics), click the gear icon next to it, and select Modify Index. There you can adjust the thresholds to your liking on a global level. Keep in mind "repeat times" determines how often the threshold must be met or exceeded in a row before an actual alarm will be triggered.

Individually - go to Performance Management > Monitoring Settings and select the checkboxes for the devices where you want to set individual thresholds, then Modify Index. You'll get a pop-up with folders for each of the performance indexes that are monitored across these devices. Expand the Packet Loss Rate ones, and notice you have a checkbox selected for "Global Index Settings". Uncheck this on either the upper folder level, device level, or individual interface level, to define a custom threshold for that index/device/interface that will be independant of the Global Index Settings mentioned above.

[pattap]

@jguse very useful! Thank you sir!

What about that hidden counter you have mentioned in the previous post, is there a way of viewing it from within the switch?

[jguse]

Hello,

You could check these on the AOS/ProCurve switch CLI using "walkmib ifInDiscards" or "walkmib ifOutDiscards". To identify which index refers to which interface, you can run "walkmib ifDescr".