Some comments:
The size is megabytes, not gigabytes.
If it is too small, bump it up. 16MB is the max. But I believe the memory is
pinned.
EX:
unxd:./>lsattr -El css0
...
spoolsize 14680064 Size of IP send buffer True
rpoolsize 16777216 Size of IP receive buffer True
...
These buffers are not related to vmtune. The error is for the CSS only.
Miles
>>> CABANL AT MODOT DOT NET 08-Oct-02 1:01:24 PM >>>
Thanks for your reply Michel, but the rpoolsize and spoolsize are already
set to 1G
root@scpm02 / # lsattr -El sys0
keylock normal State of system keylock at boot time
False
maxbuf 20 Maximum number of pages in block I/O BUFFER
CACHE True
maxmbuf 0 Maximum Kbytes of real memory allowed for MBUFS
True
maxuproc 10000 Maximum number of PROCESSES allowed per user
True
autorestart false Automatically REBOOT system after a crash
True
iostat false Continuously maintain DISK I/O history
True
realmem 4194304 Amount of usable physical memory in Kbytes
False
root@scpm02 / # lsattr -l css0 -E
AMCC_bus_io 0x1ec00 TB3MX user memory address
False
sram_memory 0xc8000000 TB3MX user memory address
False
sram_size 0x00800000 TB3MX user memory address
True
XILINX_Regs_io 0x1e800 TB3MX user memory address
False
int_priority 3 Interrupt priority
False
int_level 36 Bus interrupt level
False
spoolsize 10485760 Size of IP send buffer
True
rpoolsize 10485760 Size of IP receive buffer
True
adapter_status css_ready Configuration status
False
win_poolsize 83886080 Total user device memory
True
win_maxsize 16777216 Maximum per-user device memory
True
win_minsize 1048576 Minimum per-user device memory
True
window VSD AVAIL AVAIL AVAIL AVAIL Adapter window owners
True
I am really wondering if it REALLY is the size of the space, and not how
many pools there are? According to the documentation, if the maxmbuf is set
to 0, then thewall value is used, and is supposed to be set for 1G or 1/2
of real memory. It is set to 1G, which is 1/2 of the real 2G of memory in
that box. So, in theory and in accordance with what you recommend, my
values are set correctly.
Because I saw that there are two ways to change the #of buffer pools-- via
the vmtune -m number_of_memory_pools and the chdev -l sys0 -a maxbuf=????
I was thinking that it was actually the number of buffer pools and NOT the
size.
any thoughts??
lisa
Michel Bisschoff
<[email protected] To: ADSM-L AT VM.MARIST DOT
EDU
LD.GOV.AU> cc:
Sent by: "ADSM: Dist Subject: Re: IP pool buffers
below threshold in errpt for css0
Stor Manager"
<ADSM-L AT VM.MARIST DOT ED
U>
10/07/2002 05:30 PM
Please respond to
"ADSM: Dist Stor
Manager"
Hi Lisa,
This error relates to the SP switch. By default the rpoolsize and spoolsize
are to small sized for SP nodes with high throughput like an ITSM server.
The following command will tell you about the switch settings:
lsattr -El css0
and look at the values for rpoolsize and spoolsize. They should be 10485760
(for the ITSM server).
Probable I'll see 2097152.
To change them:
/usr/lpp/ssp/css/chgcss -l css0 -a spoolsize=10485760 -a
rpoolsize=10485760
As long as your SP nodes don't complain, there is no need to change these
values.
But you need a reboot to make the change effective. These values will be
remembered after a reboot, but
I noticed, sometimes they return to their default values. I think it
happens
after a PSSP upgrade.
So it'll be better to put this command in the file:
/tftpboot/tuning.cust
so it'll be executed automatically after every reboot.
HTH
Michel Bisschoff
Unix Systems Admin
Brisbane, Australia
-----Original Message-----
From: ADSM: Dist Stor Manager [mailto:ADSM-L AT VM.MARIST DOT EDU]On Behalf Of
Lisa Cabanas
Sent: Tuesday, 8 October 2002 2:59 AM
To: ADSM-L AT VM.MARIST DOT EDU
Subject: IP pool buffers below threshold in errpt for css0
Hello *,
Has anyone seen this before &/or knows what I should change (maxbuf or a
vmtune -m) to make it go away? I have been looking in the SP, PSSP, and
AIX performance books, and tried to find it on IBM's site, but no luck
other than I *think* I know what I am looking at,
i.e.,
This is telling me that there aren't enough buffers (as opposed to them not
being big enough) in the TCP/IP receive memory buffer. And that this error
is occurring for communication across the switch.
I need one of you AIX super geniuses to help me out, please!
thanks
lisa
LABEL: IP_POOL_BT_RE
IDENTIFIER: 454AAA48
Date/Time: Sun Oct 6 23:00:39
Sequence Number: 10357
Machine Id: 000F964F4C00
Node Id: scpm02
Class: S
Type: INFO
Resource Name: css
Description
IP pool buffers below threshold
Probable Causes
Not enough pool buffers for workload
User Causes
None
Recommended Actions
Pool tuning required
Detail Data
Software ID String
LPP=PSSP,Fn
=../../../../../../src/ssp/css/ip/tb3pci/if_ls.c,SID=1.18,L#=3332
POOL NAME
recvpool
CURRENT POOL SIZE
00A0 0000
---------------------------------------------------------------------------
|