Packet loss on Ethernet p2p link

[santoshdpawar]

Hi Guys,

I have two routers, Cisco 3825 and 2811 connected via a p2p ethernet link. The two sides of the links have Huawei muxes with 100/ full duplex set on their interface. I am facing almost 40% packet loss but the supplier is unable to identify any issues at their end. We have tried setting different combinations of duplex and speed settings but no improvement. When set to auto/auto the interfaces are negotiating to half duplex and I am sure this is the cause of the packet loss. Also I heard of auto-negotiation on MUX ports but supplier confirmed no such settings but only 100 Full duplex. If the router iterface is negotiating to half duplex so I suppose there is some thing with the negotition problem with MUX and router port ?    

On C3825 I have a Gig interface whileas C2811 has fastEthernet. I guess Gig interface should not negotiate ? Is there a way to pin point the exact fault to supplier ? Also is there work around for this?

Best regards,

Santosh Pawar       

[Joseph W. Doherty · Accepted solution]

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

If other side is hard coded full duplex, setting your side to auto should result in your interfaces running at half.

If other side is running at 100, your 3825 gig ports should run at the same.

Does your provider guarantee 100 Mbps across this p2p?  Often Metro Ethernet hand-offs can provision a lower logical CIR.  If such is your case, exceeding the logical cap normally results in packets exceeding the bandwidth being dropped.

[santoshdpawar]

Hi Joseph, Steve,

Thanks a lot for the replies. It is capped to 20 Meg and supplier has confirmed 100 Meg/ Full duplex end to end. I have tried setting 100 Full at both the ends and actually it is now running with the same settings.

I now feel that supplier has set 100 Full and Cisco router is trying to negotiate but not getting handshake indicating auto negotiation turned off in the access path. Actually supplier has never said anything about auto negotiation.

There are other links on these routers performing well and I am just thinking how to take this further as the supplier has taken their hands off from this fault. I have requested them to check with the Huawei TAC but I dont think they will check at that level. I will try with few more combinations at the two ends to see if gives any improvements. I guess we can turn off negotiation on gig interface (no-negoatiate on cisco 3825) ? Also let me know if there is any workaround that still can be implemented or tests/ debugs to prove supplier or interface config miss matches.

Best regards,

Santosh Pawar

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages whatsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

If your provider is hard coding 100/full, you should at least hard code duplex, generally you would, in this case, hard code speed too.

If your p2p circuit has a logical cap of 20 Mbps, you should shape on both your routers for that bandwidth.  (NB: my belief is most ISR shaping doesn't account for L2 overhead, so if I'm correct, you want to set your shaping 5 to 15% slower to allow for it.  Also, you might ask your provider what Bc or Tc they use when enforcing the 20 Mbps.)

[Steve Fuller · Accepted solution]

Hi,

When running auto-negotiation, if an interface operates at half-duplex it's more usually because it has fallen back to half-duplex rather than negotiated half-duplex. An interface will typically do this as the link partner either doesn't support auto-negotiation or it's disabled i.e., because the link partner has hard set speed and duplex.

You say you're tried "different combinations of duplex and speed setting" so I assume you've tried hard setting speed 100 and duplex full under the router interfaces?

Regards

[santoshdpawar]

Hi Guys,

There is an interesting thing here. The packet loss are just dissapearing if I turn off BGP. Any clue ? I tried searching for this and some one removed static route. But for case there is no static route and its /30 subnet on the router interfaces and BGP built over it. Is there anything with IOS or may be BGP ?

Best regards,

Santosh Pawar

[santoshdpawar]

Hi Guys,

Anyone faced such issue before ? Why there are no drops when we disable BGP ?

Best regards,

Santosh

[Joseph W. Doherty]

Disclaimer

The  Author of this posting offers the information contained within this  posting without consideration and with the reader's understanding that  there's no implied or expressed suitability or fitness for any purpose.  Information provided is for informational purposes only and should not  be construed as rendering professional advice of any kind. Usage of this  posting's information is solely at reader's own risk.

Liability Disclaimer

In  no event shall Author be liable for any damages whatsoever (including,  without limitation, damages for loss of use, data or profit) arising out  of the use or inability to use the posting's information even if Author  has been advised of the possibility of such damage.

Posting

Unless you've bumped into some very strange bug, the only reason that comes to my mind why removing BGP would reduce drops, does the effective path change for any traffic with removal of BGP?