2950 CRC ERRORS INPUT ERRORS

[scion515]

Hi,
This is my first post and hope ill get some good answers. I have w 4507's and 1 2950, the problem is that servers on 2950 have a small amount of packet loss like 2000 packets sent 1 or 2 loss. I can not have that. The only problem i could saw is that i see some CRC errors and Input errors that the server is on.


4507 City-Fiber Connection>>> +4507 fiber to 2950....


this is the current setting please help thanks ?



sh int fa0/3
FastEthernet0/3 is up, line protocol is up (connected)
Hardware is Fast Ethernet, address is 0014.a85b.8443 (bia 0014.a85b.8443)
Description:Server 20
MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec,
reliability 255/255, txload 28/255, rxload 1/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, media type is 100BaseTX
input flow-control is unsupported output flow-control is unsupported
ARP type: ARPA, ARP Timeout 04:00:00
Last input never, output 00:00:00, output hang never
Last clearing of "show interface" counters never
Input queue: 0/75/0/0 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 7000 bits/sec, 5 packets/sec
5 minute output rate 11046000 bits/sec, 980 packets/sec
32284072 packets input, 2318079625 bytes, 0 no buffer
Received 134372 broadcasts (0 multicast)
0 runts, 0 giants, 0 throttles
34 input errors, 34 CRC, 0 frame, 0 overrun, 0 ignored
0 watchdog, 123280 multicast, 0 pause input
0 input packets with dribble condition detected
179374132 packets output, 2393906010 bytes, 0 underruns
0 output errors, 0 collisions, 2 interface resets
0 babbles, 0 late collision, 0 deferred
0 lost carrier, 0 no carrier, 0 PAUSE output
0 output buffer failures, 0 output buffers swapped out

all it has on that port

switchport mode access vlan 2

thats it. and thats how it was for 1.5 year never had any problems

i even swap the 2950

 

[voltron1011]

Perhaps your server's nic has some issues. Usually when you see stuff like that it's associated with a machine's nic... It starts off slowly at first, then it gradually degrades.

 

[baddos]

Are you using media converters? If so, check with the manufacturer to be sure what port settings you should use speed and duplex.

 

[JOAMON]

32 million packets....2 trillion bytes...only 34 input and CRC errors....pretty good ratio. Those errors were probably only there because of the interface resets. Do a clear counters and see if they are persistent or were just one time issues.

 

[scion515]

Baddos not bad idea that was something that i checked before too and i saw couple of time the connection to the server has changed to Duplex full , dont know i still have this issue , i changed cords too , no mac-address on any edge devices ware failing , because if u dont put the mac-address to the table any edge device will shoot over all ports until the router doesent find the correct mac.....so thats not it.

I check also the spee gig fiber speed between the 2 L3's 4507 routers in 2 cities they are set to speed none which it works just fine. So its kind of wierd that the ping let say

ping x.x.x.x -n 4000 -l 30000 i get for example 3 lost which is 99% or 100% lost packets on average, but guess what i cant have that nor was that way before. What i notice is every minute or two either ping drops 1 packet or goes from 10-15ms to 500-600ms ....any reasons?

i got 5 servers on it all do the same, but none of them loose the packet on the other side of a gig link "one net transport"

 

[baddos]

What does your uplink ports on your 2950 look like?

 

[scion515]

What do you mean by that?

 

[baddos]

You have gigabit ports that connect to your 4507. What do their interfaces look like?

 

[jneiberger]

I wouldn't even waste time troubleshooting this problem over a frame error ratio that low. Thirty four errors in over 32,000,000 frames is so ridiculously low that it would be pointless to pursue. That's like a 0.00005 percent error rate or something like that. I think I slipped a decimal place the point remains. That's virtually a zero percent error rate. Why bother even looking into it?

 

[JOAMON]

I agree. Also if you notice the 2 interface resets as these are probably what made those CRC and INPUT errors anyway.

 

[baddos]

His problem isn't that he is seeing those error counters, it's the packetloss he is seeing on his servers.

I'm thinking there might be a bandwidth limitation somewhere down the road where he may need to move up to gigabit or etherchannel.

 

[jneiberger]

How are you determining that packet loss is occurring, and at what rate?

 

[scion515]

Ok

its not CRC's or Errors that i have problems with, as it looks like the CRC's are just something that came up when i configured that switch year ago.

The problem i have with is, that i loose packet's on 4 servers.

How i am determinating packet loss?

I state that on beginingin i am sending 4000-5000 pings with 30000b which it should not drop not 1 packet. Because i am moving in between 10-20mbps on other "sites" so as this one. For some reason 3-4 weeks ago that problem occur that i can't realy tell what the problem is. It might be the server it self but 4 of them?

All ports on the servers are gig'e , only the switch is set to duplex full speed 100 sin'ce it can not handle more then that.

Anyways,

The wierd thing is when i login to 4507 switch that is at the same site i don't have any problems with commands to enter etc. Once i get to that freaky 2950 my connection slow down. So my point i think is that i have some problem right there.

Here is some of the config i am running right now. Sorry i can't give you all info.

spanning-tree mode pvst
no spanning-tree optimize bpdu transmission
spanning-tree extend system-id
no spanning-tree vlan 1-2,10

interface FastEthernet0/3
descriptionServer 20
switchport access vlan 2

interface FastEthernet0/12
description 2950 to 4507
switchport access vlan 2

interface Vlan2
ip address x.x.x.x 255.255.255.0
no ip route-cache


Interface from 2950 to 4507


FastEthernet0/12 is up, line protocol is up (connected)
Hardware is Fast Ethernet, address is MAC.MAC.MAC.MAC(bia MAC.MAC.MAC.MAC)
Description: 2950 to 4507
MTU 1500 bytes, BW 100000 Kbit, DLY 100 usec,
reliability 255/255, txload 1/255, rxload 30/255
Encapsulation ARPA, loopback not set
Keepalive set (10 sec)
Full-duplex, 100Mb/s, media type is 100BaseTX
input flow-control is unsupported output flow-control is unsupported
ARP type: ARPA, ARP Timeout 04:00:00
Last input 00:00:00, output 00:00:01, output hang never
Last clearing of "show interface" counters never
Input queue: 0/75/553681/0 (size/max/drops/flushes); Total output drops: 0
Queueing strategy: fifo
Output queue: 0/40 (size/max)
5 minute input rate 12154000 bits/sec, 1109 packets/sec
5 minute output rate 6000 bits/sec, 5 packets/sec
1029478331 packets input, 3009455488 bytes, 0 no buffer
Received 867287488 broadcasts (0 multicast)
0 runts, 0 giants, 0 throttles
0 input errors, 0 CRC, 0 frame, 0 overrun, 50613 ignored
0 watchdog, 859386696 multicast, 0 pause input
0 input packets with dribble condition detected
83787000 packets output, 2695597264 bytes, 0 underruns
0 output errors, 0 collisions, 3 interface resets
0 babbles, 0 late collision, 0 deferred
0 lost carrier, 0 no carrier, 0 PAUSE output
0 output buffer failures, 0 output buffers swapped out

CPU utilization for five seconds: 7%/3%; one minute: 12%; five minutes: 19%

I think i have broadcast storm

 

[jneiberger]

It certainly appears that way. Over 80% of your traffic is broadcasts. Using this output as a reference, that amounts to over 800 packets/second of broadcasts.

Do you have a way to set up a span port and grab some of the traffic output with a sniffer? If you don't have a sniffer, ethereal is a good free tool that you can use. You should be able to grab the source IP address and/or source MAC address of the offender(s).

 

[scion515]

i do and i did, but not from the switch it self . I should try that

Forgot about doing e.real

Thanks !

How do i shoot the bcast?

 

[jneiberger]

I'm not sure what you mean by "How do I shoot the bcast?" If you span that switch traffic over to a sniffer you should very quickly be able to determine at least who the offender is and possibly even what is going on depending on the types of broadcasts you see. It could be NetBIOS, ARP, or a bunch of other things. Determining what kind of broadcast you're seeing and where it's coming from should be all you need to fix the problem.

If nothing else, you can track down the MAC address of the offender and shutdown the switch port.

 

[baddos]

Well... You don't need to setup a SPAN port to monitor broadcast traffic.


One thing that concerns me is this statement:

no spanning-tree vlan 1-2,10

Why do you have spanning-tree disabled on vlan 2 (the vlan your 2950 is setup for). Maybe there is a loop in your network.

The packetloss you are seeing is definately coming from this overloaded interface because of a misconfiguration somewhere on vlan2.

 

[jneiberger]

heh heh....good point about the broadcast traffic and the span port. ;-) It is a little pointless to setup a span port to see broadcast frames.

I guess I need to go get some more coffee.

John

 

[scion515]

So what you are saying is that i should remove the vl 2 from spann?

 

[jneiberger]

You already have STP disabled on VLAN 2. I think the poster is recommending that you turn it back on. Do you have a good reason for disabling STP?