3750 Receive CRC Errors

Unanswered Question
Jul 27th, 2010

I have a 3750G-PoE-48 stack with 4 units.  Member 3 of the stack has around 15 ports with Receive Errors ranging from under 100 to almost 50,000.  None of the other stack members show any Receive Errors.  Could it just be coincident on the devices plugged in to this switch are at issue or could Member 3 have some issues.  Not sure where to begin troubleshooting.


  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Siddharth Chand... Tue, 07/27/2010 - 12:44


What devices are connected to these 15 ports on the 3750g switch ?

Please paste the output of " show interface gig <> "  and " show run int <> " for a couple of ports.

Are you also seeing CRC errors along with rcv-err ?

- Sid

fasteddye Tue, 07/27/2010 - 13:06

Fairly new HP Computers are on these ports.  I believe I am seeing CRC errors as well as Receive Errors.

I have attached the sh int gig and sh run int of 2 ports.


Siddharth Chand... Tue, 07/27/2010 - 13:21


     395773 input errors, 395773 CRC, 0 frame, 0 overrun, 0 ignored

All the errors are due to CRC errors.

Causes for this:

1. Bad port 3750.

2. Bad cable.

3. Bad frames sent by the device on the other side.

4. NIC card imcompatibilty between HP and the Cisco switch (unlikely)

Take a test port on the 3750 switch that shows CRC errors.

- Change the cable. - Check

- Connect a different brand test p.c (diff NIC or diff manufact) on the same port, see if the CRC errors still increment.

   If no, its a bad port, nic, incompatibilty issue with the original HP p.c

   If yes, then connect the same p.c on a different 3750 switch, if CRC errors dont increment, then the switchport on orignal 3750 is bad.

Its a issue of narrowing down causes by testing.

- Sid

fasteddye Tue, 07/27/2010 - 13:53

Not sure if this is helpful but on the web interface these ports will show orange sometimes and say "port has faulty link".

Thanks for the suggestions.

Leo Laohoo Tue, 07/27/2010 - 14:48

1.  Are the clients NIC cards set for auto-negotiation of speed and duplex?  Recommend that you enable this.

2.  Can you post the output of the command "sh controller ethernet " of the two ports you sampled?

3.  To test if you have a bad cable, use the command "test cable tdr interface " and wait for about 5-7 seconds and run the second command "show cable tdr interface ".  Please post the output too.

Hope this helps.

fasteddye Tue, 07/27/2010 - 19:04

I will need to verify nic settings in the morning but I would be shocked to find them not set to auto.

I ran the sho controller ethernet command.  I also ran the test cable tdr command.

The results are attached.


Leo Laohoo Tue, 07/27/2010 - 19:14

Thanks for this.  The TDR test shows all OK so that eliminates faulty cabling.

The "sh contr e", however, shows that in both interface, under "Receive", you are getting abnormally high FCS and Collision fragments.  So I'd request you check if the NICs are auto-negotiated.

fasteddye Wed, 07/28/2010 - 13:21

NICs appear to be set to auto.  Could this be a faulty switch?

Nagaraja Thanthry Wed, 07/28/2010 - 13:29


What kind of cable you are using? Can you please verify that the cable connecting the switch and the PC is CAT5e or higher? Also, if you have any patch panel along the path, can you bypass that and see if it helps.



fasteddye Wed, 07/28/2010 - 13:48

I have run the  "test cable tdr interface " command, would that not alleviate cabling?

I will have to verify cables, our standard is 5e or better but you never know.  It would not be possible to bypass patch panel.

August Ritchie Wed, 07/28/2010 - 13:52

Nagaraja is right,

I have seen patch panels be the causes of CRC errors more than a few times, if replacing the cabling doesn't work you will need to find someway to test without the patch panel.


This Discussion