Input and CRC Errors

Unanswered Question
Apr 1st, 2009

Hi All,

We have been having a very strange problem affecting a lot of users within our company and we are struggling to understand the underlying issue.

Briefly, we have a range of switches (but mainly 2950's and 3750's) with various levels of IOS located in the UK. We are getting consisent reports of slow responses when people log in using windows which seems to be the roaming profile being downloaded from the local DC.

On further investigation we have found that the DC's (mainly Windows 2003) are erroring on the port with a large amount of Input and CRC errors. However once we reset the port (either shut/no shut) or a physical pull of the cable then the errors disappear and all is ok.

The monitoring we have shows no real problems beforehand and no correlation between these problems. The other thing is that this is a mixture of machines and network cards , Dell/HP/Broadcom etc.

We are stuck by this and we can not understand any correlation or single issue causing this.

For servers we hard code to port speed of server.

Cheers,

Ian

I have this problem too.
0 votes
  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
thotsaphon Wed, 04/01/2009 - 08:10

Lan,

I should check the cable and the duplex on both ends real quick when I got CRC errors.

Edit: John provided you with good information. Just to confirm what we are thinking of

###########################

CRC

Description: Cisco IOS sh interfaces counter. This increments when the CRC generated by the originating LAN station or far-end device does not match the checksum calculated from the data received.

Common Causes: This usually indicates noise or transmission problems on the LAN interface or the LAN itself. A high number of CRCs is usually the result of collisions but can also indicate a physical issue (such as cabling, bad interface or NIC) or a duplex mismatch.

###########################

BTW, John How are you doing?

HTH,

Toshi

ian.robathan Wed, 04/01/2009 - 08:29

Thanks Toshi, however whilst I can accept one or two may be cable problems, we have had over 70 servers in different locations affected by this and a simple shut/no shut resolves, iif it was cable, loose connection etc. I am not sure how this would resolve it.

Cheers,

Ian

John Blakley Wed, 04/01/2009 - 08:10

Ian,

I would hard code the speeds on the ports that the servers need to be at. Generally, CRC errors indicate a physical problem (mainly duplex). Try setting them 1000/full (or 100/full if they're 10/100 ports) for the port that the servers connect to and see if that helps.

HTH,

John

ian.robathan Wed, 04/01/2009 - 08:27

thanks John, I just actually added we do that as you posted. All servers are hardcoded, PC's for we get the problem on them as well are set to Auto as were advised to by Cisco.

larry.atkins Wed, 04/01/2009 - 08:30

Make sure you set the switch ports to the proper speed/duplex also. The errors could also be cause by bad cables.

Leo Laohoo Wed, 04/01/2009 - 14:10

We've had similar issues like this. In our network environment, we have a number of servers and some of them are old DL380's. We discovered that if talking with "dual speed" NICs (10/100BaseT) from a "triple speed" NICs (10/100/1000BaseT), it's better to hardcode the switch to 100/Full on both sides. This has helped us alot. However, if the server's NIC and the switch is triple speed, then auto negotiate is OK.

Hope this helps.

ian.robathan Thu, 04/02/2009 - 00:29

Hi Leo, as a standard we set the speed on switch and servers to 100/full, this has been the case across all our different types of switches/servers and for the majority it is OK. Because the problem is random we seem to have no particular pattern.

Cheers,

Ian

Actions

This Discussion