Mr. Joshua,
This looks like a bug I'm working with:
"Anyway, the issue might be related to the middle buffer "not populating"
and there is a bug which was open for the similar issue
(should be fixed in later IOS versions): CSCdx15857 (Buffer Failure). You
couldn't change the middle buffers because of the same bug.
In order to fix the issue you should upgrade the IOS but prior of doing this
you would probably need to consult your Cisco NSA/SE."
That came from a TAC guy I'm working with. He's telling me the 12.1.13E
code will fix that problem, where you can't even configure medium buffers.
Keep in mind that buffers use RAM, so occasionally do a 'sh mem' and make
sure your 2 pools aren't running low. I'd install that code, and then run
it for a few days. After that, set your permanent buffers to between 50 and
75% of what the peak was for that particular pool. Set the max to maybe 100
more than the permanent. So for below I'd start out with:
buff sma per 750
buff sma max 850
buff med per 75
buff med max 150
buff mid per 300
buff mid max 400
buff big per 600
buff big max 700
buff very per 20
(use default for very max)
buff large per 10
buff large max 25
buff huge per 10
buff huge max 20
Paste these in, see how it goes. Good luck.
Chuck Church
CCIE #8776, MCNE, MCSE
Sr. Network Engineer
Magnacom Technologies
140 N. Rt. 303
Valley Cottage, NY 10989
845-267-4000
Date: Tue, 5 Nov 2002 12:41:12 GMT
From: "Mr Joshua" >
Subject: buffer tuning 6509 [7:56891]
Does anybody know of a good white paper on buffer tuning? I have read a
couple of generic ones on Cisco's website, yet they are not
good enough to satisfy what I need to know. Called TAC - not a whole
lot of help this time! As you can see, there are a lot of misses on
medium and middle buffers. I also see that total and permanent are
not allocated. I know the general CCNP level of what those mean and
commands to adjust them, but does anybody know this - the second line
of output says that there are 500 max allowed. Does that mean that
I need to break this number down into public buffer pool? Does that mean
that the cumulative sum of all public pools can't be more than 500? (as you
can see, the big buffers are 500). Does anybody know of a GOOD paper that
gives examples of buffer tuning? Sorry if those are stupid questions.
here is the output:
Buffer elements:
499 in free list (500 max allowed)
898918875 hits, 0 misses, 0 created
Public buffer pools:
Small buffers, 104 bytes (total 73, permanent 50, peak 1501 @ 7w0d):
72 in free list (20 min, 150 max allowed)
609248534 hits, 201320 misses, 121659 trims, 121682 created
86630 failures (0 no memory)
Medium buffers, 256 bytes (total 0, permanent 0, peak 123 @ 4d08h):
0 in free list (0 min, 0 max allowed)
705511 hits, 140644897 misses, 1414484 trims, 1414484 created
139937655 failures (0 no memory)
Middle buffers, 600 bytes (total 150, permanent 25, peak 555 @ 7w0d):
149 in free list (10 min, 150 max allowed)
185320811 hits, 4615702 misses, 167032 trims, 167157 created
4439672 failures (0 no memory)
Big buffers, 1524 bytes (total 500, permanent 500, peak 595 @ 7w0d):
500 in free list (5 min, 500 max allowed)
41418467 hits, 3577401 misses, 39229 trims, 39229 created
3540388 failures (0 no memory)
VeryBig buffers, 4520 bytes (total 10, permanent 10, peak 20 @ 7w0d):
10 in free list (0 min, 100 max allowed)
1006090 hits, 3524469 misses, 22 trims, 22 created
3524458 failures (0 no memory)
Large buffers, 5024 bytes (total 0, permanent 0):
0 in free list (0 min, 10 max allowed)
0 hits, 3524458 misses, 0 trims, 0 created
3524458 failures (0 no memory)
Huge buffers, 18024 bytes (total 2, permanent 0, peak 2 @ 7w0d):
2 in free list (0 min, 4 max allowed)
4580 hits, 3522061 misses, 120 trims, 122 created
3522000 failures (0 no memory)
Message Posted at:
http://www.groupstudy.com/form/read.php?f=7&i=56906&t=56906
--------------------------------------------------
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
