With 12.1(8b)E9 be aware of this bug,
CSCdw89942 <http://www.cisco.com/cgi-bin/bugtool/onebug.pl?bugid=CSCdw89942>
SH
-----Original Message-----
From: Steve Francis [mailto:sfrancis@expertcity.com]
Sent: Wednesday, May 22, 2002 2:55 PM
To: Stephen Sprunk
Cc: cisco-nsp@puck.nether.net
Subject: [nsp] Which IOS? (was Re: Priority queue)
So how does a person select the most stable IOS image at any given time?
I typically deploy the latest release in the most conservative code train
that supports the hardware/features.
However, that has me running the now deferred
12.1(11b)E2
on my MSFC2's that I recently deployed.
You recommend
12.1(8b)E9
.
Another cisco employee recommended that I use "12.1(8b)E10. Optionally,
12.1(11b)E3 which was just released." as the most stable code possible.
Does cisco collect stats as to the number of deployments of different code
levels, and problems found therein, and the length of uptime?
How is one to make an informed guess?
How does everyone else do it?
TIA
Stephen Sprunk wrote:
Thus spake "Felix Lee" <mailto:felixlee_hk@hotmail.com>
<felixlee_hk@hotmail.com>
I have a Catalyst 6509 running with Fw 5.4(2) & NmpSW 5.5(11). And the
MSFC is running on 12.1(3a)E4.
FWIW, you really don't want to be running 12.1(3a)E4. We're having very
good luck with 12.1(8b)E9 so far. 5.5(11) is okay but I'd go to 5.5(13a)
when convenient.
One strange point that cannot be explained is the priority
queue applied on a VLAN which is connection to a WAN
100Mbps fastethernet. The priority queue is to put the
FTP traffic low priority. After applying the priority queue,
the time for a daily FTP job has been extended from 3
hours to 5 hours while the traffic loading of the link is only
around 10%. The duration come back to 3 hours after
removing the priority queue.
Can someone help to give advice?
When you put traffic in a low priority queue, that's going to increase
latency to some degree. TCP throughput is determined by bandwidth*delay.
Therefore, putting FTP in the low queue is going to slow it down.
I wouldn't expect such a drastic difference with only 10% loading, but if
you look closer, you're probably seeing periodic spikes which only average
out to 10%. These spikes will hammer your FTP's congestion avoidance
algorithm if it's the only significant activity in the low queue.
S
This archive was generated by hypermail 2b29 : Sun Aug 04 2002 - 04:13:45 EDT