cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
16905
Views
5
Helpful
31
Replies

Port Disables Every 10 Minutes

TrustNetworks
Level 1
Level 1

Good Afternoon

I am experiencing the same problem described in this post https://supportforums.cisco.com/message/3421454#3421454

I have seen this happen on different networks, with different equipment attached. It happens on both 2960 and 3750 switches.

Basically the connection drops, and we see in the web interface "Port is Disabled". This appears to happen every 10 minutes.

On the CLI, the status shows as connected.

Port      Name               Status       Vlan       Duplex  Speed Type

Fa0/38                       connected    1          a-full  a-100 10/100BaseTX

I have ran cable diagnostics while the drop out is occuring.

Interface Speed Local pair Pair length        Remote pair Pair status

--------- ----- ---------- ------------------ ----------- --------------------

Fa0/38    100M  Pair A     28   +/- 15 meters Pair B      Normal

                        Pair B     28   +/- 15 meters Pair A      Normal

                        Pair C     N/A                Pair D      Not Supported

                        Pair D     N/A                Pair C      Not Supported

During the outage, I see the duplex flucate between full and half. The outage occurs for approx 90 seconds

If I fix the duplex and speed at both ends, the outage reduces to around 30 seconds.

If I apply spanning-tree portfast the outage reduces further to around 10 seconds.

Before I change any configuration on the port, the logs show the interface going down

Aug 16 13:06:51.875: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/38, changed state to down

Aug 16 13:06:52.874: %LINK-3-UPDOWN: Interface FastEthernet0/38, changed state to down

Aug 16 13:06:55.306: %LINK-3-UPDOWN: Interface FastEthernet0/38, changed state to up

Aug 16 13:06:56.313: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/38, changed state to up

Aug 16 13:07:28.089: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/38, changed state to down

Aug 16 13:07:30.094: %LINEPROTO-5-UPDOWN: Line protocol on Interface FastEthernet0/38, changed state to up

However, once I apply the configuration nothing is logged. However we can still see the conneciton is disappearing for around 10 seconds.

I suspect the issue wasn't resolved for the person reporting the problem in the link above, but because the outage is minimized, and not being logged it is going unnoticed.

Does anyone know what causes this, or how I can troubleshoot further.

Regards

Paul

31 Replies 31

Leo Laohoo
Hall of Fame
Hall of Fame

What's connected to this port?

Judging from what you are describing, I'm suspecting the client connected to port F 0/38 has a faulty NIC or an improperly configured NIC.

Can you post the output to the following commands:

1.  sh version;

2.  sh interface F0/38; and

3.  sh controll e F0/38

Hi Leolaohoo

That was my first thought, however we can see this happning across a variety of devices. Connected to this particualr port is a Canon ir3235 Printer. However we have seen it happen with Dell Servers, Dell and HP Workstations, Sony Vaio laptops.

I also see someone reporting what appears to be a similar issue on expert sexchange, relating to IP Cameras. (Link: http://www.experts-exchange.com/Hardware/Networking_Hardware/Switches/Q_27639662.html)

I'm not 100% but think it may only be happening on switches that are supporting more than 1 VLAN, with devices on the native VLAN. I have tried disabling STP, with no change.

Requested Output attached.

That was my first thought, however we can see this happning across a variety of devices. Connected to this particualr port is a Canon ir3235 Printer. However we have seen it happen with Dell Servers, Dell and HP Workstations, Sony Vaio laptops.

Can you elaborate further?  How are these devices connected to the switch?

Are these devices connected to the same switch?  Using the same patch cord?  Using the same patch port?

Can you try by connecting anyone one of these devices DIRECTLY to the switch?

Everything is patched directly into 2 x 48 port switches.

Everything is patched directly into 2 x 48 port switches.

So let me get this straight:  The cabling is from the client < -- > ethernet cable < -- > switch?

Is this correct?

There are not patch paneling in between?

The equipment is patched through a passive cabling system, this won't be the problem though.

Sent from Cisco Technical Support iPhone App

The equipment is patched through a passive cabling system, this won't be the problem though.

I agree with Mike here.  You cannot be sure that your patch cabling system is foolproof.

Here's one way of testing:  Directly connect the client to the switch.  Clear the interface counters (clear counters F0/BLAH) and the controller (clear controll e f0/BLAH).

Test for a few days.  If you don't see the interface drop or bounce then you know who to call.

NOTE:  We have a Fluke DTX-1800 TDR with us.  Our cable contractors are "scared" when we tell them that we're going to verify a problematic line using this. 


Hi Leo, Mike

Dont get me wrong, I'm grateful for your input.  Your correct in a can not absolutley 100% guarantee the cabling system isnt fool proof.

However some of the cabling has been done by us, and is new. Some of the cabling was pre-installed by our customers previous provider. Certainly with the cabling installed by us, this has been tested off with a fluke scanner, as required to provide the manufacturers warranty.

This particular customer had no issues before we installed the new Cisco switches, while I would agree this would be a useful test, and if the oppotunity arises I will carry out this test. However as we look after the customer remotley, and 2 x site visits and leaving a cable strewn across the customers office is both has an associated cost and far from ideal.

I will also re-iterate this is not the first time I have seen this issue, I have seen it previously and it is always with Cisco switches.

As im as confident as I can be the cabling system, also note the cable test posted in my orginal post done at the switch end.

I would be grateful if anyone has either seen this problem before, has any other input, or can advise of any further tests that can be carried out to further identify this issue.

Just to add, I have recently discovered that some workstations are dropping every 15 minutes rather than 10 minutes. My feeling here is that there is likley to be some compatabillity issue with something either configured on the switch or tthe switch itself, and particluar 'types' of network adaptors.

Regards

Paul

Just to add, I have recently discovered that some workstations are dropping every 15 minutes rather than 10 minutes. My feeling here is that there is likley to be some compatabillity issue with something either configured on the switch or tthe switch itself, and particluar 'types' of network adaptors.

I've seen something like this happened a few years ago.  But only WORST.  Instead of "dropping" every 15 minutes, the interfaces were dropping like >12 times per minute.  This was due to the fact that my client refused to upgrade their 3500XL switches when they were introducing brand-new computers.  The Intel NIC cards were incompatible to the very, very old 3500XL.

Regarding the cabling "issues", can you directly connect the client to the switch.  Clear the interface counters (clear counters F0/BLAH) and the controller (clear controll e f0/BLAH).  Let's observe if this issue re-occurs.


This is very strange because I have >300 2960/2960S (and >100 3750/G/E/X too) in my network and I've never seen a series of behaviours like yours. 

Can you also specify what IOS are your fleet of switches running on?

Test for a few days.  If you don't see the interface drop or bounce then you know who to call.

Mert OZKUL
Level 1
Level 1

Hi,

Maybe not solves your problem ( i'm really interested what will going to experts says about that too ) but are you using EEM to handle that problem automatically temporarily?

or what is your workaround until the problem solved?

BR

Hi BR

I'm not using EEM, as these are printers, workstations etc. The only work-around I have is to set all ports to portfast for STP, this minimises the outage to approx 10 seconds, however not ideal as a couple of customers have inadvertently created loops before.

Sent from Cisco Technical Support iPhone App

What is your logic that the patch cabling system won't be the problem? We had a contractor that came in and installed patch panels and we had all kinds of duplex issues. They finally came in and re-punched everything and problem solved. You can only rule that out if you connect something directly into the switch and the issue still happens, in fact you should do that and see what happens.

Mike

TrustNetworks
Level 1
Level 1

Mainly because we have seen it happen at different customers sites (including servers plugged directly to the switch).

Also because the problem occurs at timed intervals on a per port basis, i.e the devices don't drop at the same time.

Sent from Cisco Technical Support iPhone App

Paul....no worries.

I re-read the posts and I did not see if you mentioned if the ports are seeing any errors increase. Can you clear the counters and check on some of the interfaces and see if any errors pop up. Do some machines have this happen at a greater frequency than some? Also can you post the config of one of the switches that is having the issue along with the show version. Also can you or have you contacted TAC?

Mike

Getting Started

Find answers to your questions by entering keywords or phrases in the Search bar above. New here? Use these resources to familiarize yourself with the community:

Innovations in Cisco Full Stack Observability - A new webinar from Cisco