What Causes Deferred packets?
Eric Arnold
of the ports on the switch is connected to Eth0/0 on the 3620. The switch is
hardcoded to 10 Mbps half-duplex, so I know there are no auto-negotiate
issues. What we are seeing is a large number of deferred packets, 20% - 30%
of the total packets between the switch and the router are being deferred.
We see these deferrals on the switch interface and the router interface. The
switch has 5 other ports active, all connected to servers running MS SNA
Server acting as SNA gateways. They are all running at 100 Mbps/full,
hardcoded on both sides as well. We're also seeing about 15% collisions, but
I'm not too worried about that, sinc ethe link is half-duplex. And no, we
apparently can't make a 10 Mbps Ethernet interface full-duplex, at least on
the router side we can't.
No other errors are being seeen, no CRC, etc. Just lots of deferrals.
Can't seem to find any decent info on Cisco's website, best they say is:
"Deferred indicates that the chip had to defer while ready to transmit a
frame because the carrier was asserted."
Sorry, I need a bit more explaination than that. Any ideas or thoughts?
Thanks
--
Eric Arnold
(remove the "_no_spam" to reply directly)
Frans Brinkman
it simply means that the interface wanted to send a frame,
found the cable buzy because the other side was sending
a packet, so it had to wait for that to finish. This is not
an error.
Hansang Bae
>Trying to troubleshoot connectivity between a 3620 and a single 2924-XL. One
>of the ports on the switch is connected to Eth0/0 on the 3620. The switch is
>hardcoded to 10 Mbps half-duplex, so I know there are no auto-negotiate
>issues. What we are seeing is a large number of deferred packets, 20% - 30%
>of the total packets between the switch and the router are being deferred.
>We see these deferrals on the switch interface and the router interface. The
>switch has 5 other ports active, all connected to servers running MS SNA
>Server acting as SNA gateways. They are all running at 100 Mbps/full,
>hardcoded on both sides as well. We're also seeing about 15% collisions, but
>I'm not too worried about that, sinc ethe link is half-duplex. And no, we
>apparently can't make a 10 Mbps Ethernet interface full-duplex, at least on
>the router side we can't.
>No other errors are being seeen, no CRC, etc. Just lots of deferrals.
>Can't seem to find any decent info on Cisco's website, best they say is:
>"Deferred indicates that the chip had to defer while ready to transmit a
>frame because the carrier was asserted."
>Sorry, I need a bit more explaination than that. Any ideas or thoughts?
No worries. In a half-duplex world, the stations have to listen to the
wire and transmit when the wire is free. That would be the Carrier
Sense part of CSMA/CD.
So the router or the switch tried to send something, saw that the wire
was busy and deferred the transmission.
hsb
"Somehow I imagined this experience would be more rewarding" Calvin
********************************************************************
Due to the volume of email that I receive, I may not not be able to
reply to emails sent to my account. Please post a followup instead.
********************************************************************
Dave Phelps
something out. You can do full duplex on the 3600s. You need somewhere around
IOS v12.0. I upgraded the software on a 3640 specifically for this ability to
v12.04.
On Fri, 6 Oct 2000 12:03:12 -0700, "Eric Arnold" <loki...@nospam.pacbell.net>
wrote:
>Trying to troubleshoot connectivity between a 3620 and a single 2924-XL. One
>of the ports on the switch is connected to Eth0/0 on the 3620. The switch is
>hardcoded to 10 Mbps half-duplex, so I know there are no auto-negotiate
>issues. What we are seeing is a large number of deferred packets, 20% - 30%
>of the total packets between the switch and the router are being deferred.
>We see these deferrals on the switch interface and the router interface. The
>switch has 5 other ports active, all connected to servers running MS SNA
>Server acting as SNA gateways. They are all running at 100 Mbps/full,
>hardcoded on both sides as well. We're also seeing about 15% collisions, but
>I'm not too worried about that, sinc ethe link is half-duplex. And no, we
>apparently can't make a 10 Mbps Ethernet interface full-duplex, at least on
>the router side we can't.
>
>No other errors are being seeen, no CRC, etc. Just lots of deferrals.
>
>Can't seem to find any decent info on Cisco's website, best they say is:
>
>"Deferred indicates that the chip had to defer while ready to transmit a
>frame because the carrier was asserted."
>
>Sorry, I need a bit more explaination than that. Any ideas or thoughts?
>
>Thanks
Dave Phelps
Phone Masters Ltd.
tippe...@nospam.com
nospam=bigfoot
Herbert Haeupler
> Trying to troubleshoot connectivity between a 3620 and a single 2924-XL. One
> of the ports on the switch is connected to Eth0/0 on the 3620. The switch is
> hardcoded to 10 Mbps half-duplex, so I know there are no auto-negotiate
> issues. What we are seeing is a large number of deferred packets, 20% - 30%
> of the total packets between the switch and the router are being deferred.
a lot too much. The router is unable to place the packets on the line because
there is too much traffic coming in (broadcasts i think).
>
> We see these deferrals on the switch interface and the router interface. The
> switch has 5 other ports active, all connected to servers running MS SNA
> Server acting as SNA gateways. They are all running at 100 Mbps/full,
> hardcoded on both sides as well. We're also seeing about 15% collisions, but
> I'm not too worried about that, sinc ethe link is half-duplex. And no, we
i would be worried about that. 15% is definitely bad! Look for an IOS version
that is capable to run the link full-duplex ( i think some 12.0's are) or spend
the
money for a fasth-ethernet nm.
>
> apparently can't make a 10 Mbps Ethernet interface full-duplex, at least on
> the router side we can't.
>
> No other errors are being seeen, no CRC, etc. Just lots of deferrals.
>
> Can't seem to find any decent info on Cisco's website, best they say is:
>
> "Deferred indicates that the chip had to defer while ready to transmit a
> frame because the carrier was asserted."
> .....
>
> --
> Eric Arnold
> (remove the "_no_spam" to reply directly)
- Herbert
Herbert E. Haeupler
Siemens I&C Networks
Nuremberg - Germany