Excessive broadcasts and CRC/Alignment errors on ProCurve 2610A

[arigsby]

I've got a ProCurve 2610A switch configured setup as the core switch on my LAN.  There are other 2610As and Cisco 2960 switches hanging off of it.  Normally I would not have end users hanging directly off my core but I'm fighting a mini-battle right now to get additional ProCurves for end users.  There's only a small handful of end users on the ProCurve, fortunately.

A few people have complained about "network slowness" and, personally, I am not seeing anything personally that would corroborate that claim.  However, the people saying this the most are folks within the IT department, itself, and include the CIO.  Further, over the weekend one of my co-workers VPN'd in and could not communicate with his blade server (which is on one of the ports I am seeing these errors for) and yet mysteriously the problem went away and everything was fine.  Keep in mind that it was preventing him from initially getting in to the blade and not the result of him "tinkering around".

I am seeing alerts within the GUI's Status tab related to

• Excessive broadcasts detected on port X
• Excessive CRC/Alignment errors on port X

Well over half of the ports are active on the core switch and almost all of the alerts are related to 5 specific ports, with a 2 of them more common than the other 3.  All but one of these machines are desktops/laptops that get their IP via DHCP.  The remaining port is used for a Blade server with static IP.  Using my Fluke, I've tested the cables everywhere from the end user's patch cable, the cabe from the wall jack to the patch panel, and from the patch panel to the switch.  I am not seeing any problems and none of these have lengths exceeding 220 ft..  N

I do have Spanning Tree enabled on the core and yes, I do have some VLANs, which are working fine.  Additionally, within the HP PCM application (which I just installed) I am not seeing any smoking guns that would indicate what the problem might be.  Most everything is green (yes, yes.... highly scientific, I know).  In the attached config, you will see that the alert sensitivity is set to high, which is what HP recommended and set the default to.  I am now wondering if there really is no serious issue to be concerned with and that the threshold for the alert is simply set too low.

Any ideas or thoughts would be appreciated.  Switch config is attached.

Thank you.

[ernestdavised]

CRC usually is a duplex mismatch...try updating the drivers or manually setting duplex speed settings.

[Richard Brodie_1]

I would be a little sceptical of what the fault-finder says, especially when it's set to high. I would definitely want to look at the interface counters, or preferably have them logged periodically before I took them too seriously.

Broadcast could easily be exceeeding a percentage, on a very low traffic volume, so of no consequence at all. CRC/alignment is harder to dismiss - although if you get a wobble just after the link comes up, it's probably not worth worrying about.

[hello007]

Excessive broadcast and multicast traffic can seriously degrade network performance. By following the configuration examples given here and disabling IP-directed broadcasts
on edge routers, it is possible to mitigate the effects of excessive broadcast and multicast traffic.

[dhoenel · Accepted solution]

Hello,

i hat this Problem too.

https://support.hpe.com/hpesc/public/docDisplay?docLocale=en_US&docId=c02597187

My solution.

I change the Patchcable on the Uplink Port of the Switch and change the transceiver too.

And it works fine.

Moderator Note: [Updated with new links]