This discussion has been locked. The information referenced herein may be inaccurate due to age, software updates, or external references.

You can no longer post new replies to this discussion. If you have a similar question you can start a new discussion in this forum.

In/Out Errors and Discards

ajdibble over 20 years ago

Hi everyone,

Can anyone give me tips on what Receive/Transmit Errors and Discards really mean? I'm trying to use the built-in Top-10 Errors Today summary chart to go through our network and fix problems. I managed to fix some Transmit Discards by nailing down the autonegotiate settings on a switch and a server - whether the discards were due to collisions or a duplex mismatch I'm not sure. At present I have an interesting one, Transmit Errors looking like this:
_ _ _
|| || ||
_||__||__||__ etc. (in case anyone can't work out what that's supposed to represent, the graph looks like a square wave) it's almost completely consistent and has been ever since I installed SolarWinds a month or so ago, doesn't seem to relate to traffic levels or anything. The machine is an NT4 BDC with a DEC 21142 Ethernet card connected to a DEC VNswitch 900CG.

Any help on the specific issue and more general tips as well would be gratefully appreciated.

Thanks,

Anthony.

Anthony

0 Brad over 20 years ago

Hi, Anthony

Auto Negotiation is a constant top issue on most networks, on our Cisco and extreme switches where servers or routers are not aligned with duplex settings, errors will occur(collisions, Rx, and Tx errors). The type of errors will depend on the traffic, example: server auto to half duplex 100mg while switch settled on full dup. 100mg, server shows more receive errors than tx errors due to inbound heavy traffic loads. I usually fix the switch at Full dup 100 and stay with auto on the switch, the errors go away. Our extreme switches have had a known issue with Intel Pro cards. In the old days of 10Meg half duplex hub, the CSMA/CD is the way things worked: with errors, collisions and discards a way of life.
Hope that helps some
Brad
Cancel
Vote Up 0 Vote Down

Cancel

0 scottsams over 20 years ago

Hey Brad,
Have you checked your port? If you set your server to full/100 and your switch to auto, I am pretty sure you will get half/100 and subpar performance. This is documented with Cisco and is a feature of the autonegotiation protocol.

Scott

quote:Originally posted by Brad

Hi, Anthony

Auto Negotiation is a constant top issue on most networks, on our Cisco and extreme switches where servers or routers are not aligned with duplex settings, errors will occur(collisions, Rx, and Tx errors). The type of errors will depend on the traffic, example: server auto to half duplex 100mg while switch settled on full dup. 100mg, server shows more receive errors than tx errors due to inbound heavy traffic loads. I usually fix the switch at Full dup 100 and stay with auto on the switch, the errors go away. Our extreme switches have had a known issue with Intel Pro cards. In the old days of 10Meg half duplex hub, the CSMA/CD is the way things worked: with errors, collisions and discards a way of life.
Hope that helps some
Brad

Scott Sams

0 ajdibble over 20 years ago

Just in case anyone's interested, this problem was fixed by moving the machine in question onto a different switch port! I tried to fix it initially by playing around with speed and duplex settings, turning off autonegotiate, but this made no difference. Just for the sake of trying it, I plugged the machine into a different network port and the errors just disappeared. I haven't tried plugging the device back into the original switch port so I don't know if the port is faulty or whether it just got itself into a knot, which could feasibly have reset itself when the port went down.

Anthony
Cancel
Vote Up 0 Vote Down

Cancel
0 ajdibble over 20 years ago

Just in case anyone's interested, this problem was fixed by moving the machine in question onto a different switch port! I tried to fix it initially by playing around with speed and duplex settings, turning off autonegotiate, but this made no difference. Just for the sake of trying it, I plugged the machine into a different network port and the errors just disappeared. I haven't tried plugging the device back into the original switch port so I don't know if the port is faulty or whether it just got itself into a knot, which could feasibly have reset itself when the port went down.

Anthony
Cancel
Vote Up 0 Vote Down

Cancel