Excessive utilization of port....

[Jonathan Axford]

Hi Guys,

Came in this morning to see that one of the ports on a 2524 switch was reporting utilization of 500%

It is a fairly busy port as it connects a massive colour printer/photocopier to the network, this gets used to print some fairly large files.

Has anyone seen this before? Is it correct or is there something wrong with my PCM?

I have attached a screen shot of the dials, Broadcasts also got quite high....

[Jesus-Perez]

Disable all protocols that you don't need in the colour printer/photocopier.
But if your print jobs are big... this could be normal.

[Jonathan Axford]

Didn't think of that, i will wander over there and take a look at it today and see if there is anything running on it that shouldn;t be....

Cheers

[Steve Britt]

Jonathan,

PCM does not pick up any changes to port ifSpeed after its initial discovery, so if the port speed changed after PCM found this device (e.g. the port was connected after initial discovery) then PCM will be using the nominal port speed rather than the negotiated speed. As you pointed out in an earlier post, deleting the device from PCM and rediscovering it may help as this will flush the device from the database (and with it the old ifSpeed); the data should be correctly populated upon rediscovery.

What version of PCM are you running? If not running 2.1 with autoupdate 6 you should consider upgrading to that version.

It looks like the screenshot might not be showing what you wanted it to ...? Very low counts on port 10.100.1.45 port 2, but the picture looks truncated.

The fact that the broadcasts are high as well may mean that you do have high traffic, though the utilization count is clearly incorrect (it should never exceed 100%). If you haven't yet deleted the suspect device and rediscovered it, try that. Also update to autoupdate 6 if you haven't yet done so.

If those suggestions fail to resolve the issue, and you're inclined to investigate further, check the ifMIB counters on the port with a MIB browser as I advised in an earlier posting. Take the difference of the octet counters over the course of a minute - if the port is full duplex then add both ingress and egress counts, divide by 60 seconds to get a rate, and then divide that number by ifSpeed (double the speed if full duplex) to see what roughly what you should be seeing for utilization. Then you should probably open a support call so we can try to replicate your situation and debug it from there.

Regards,

SVB