From: Barry Tait (Barry.Tait@aggreko.co.uk)
Date: Wed Feb 08 2006 - 09:02:57 EST
All,
My problem is v440 is chronically slow when ftp'ing large files to it -
I am
seeing loads of runts on the Cisco switch between source and v440
target.
I have a new v440 which came preinstalled with Solaris 10.
I installed Solaris 8, wiping Solaris 10 out (don't ask).
# uname -a
SunOS gla1mfg1 5.8 Generic_108528-22 sun4u sparc SUNW,Sun-Fire-V440
# cat /etc/release
Solaris 8 HW 7/03 s28s_hw3wos_05a SPARC
Copyright 2003 Sun Microsystems, Inc. All Rights Reserved.
Assembled 11 June 2003
#
In 1hr 45 mins (6300 seconds), I transferred about 4.5Gb.
I need to speed up this connection to normal speed to allow job to
complete asap.
##################
I am working on a 100Mbit/s LAN:
I worked out that maximum theoretical throughput over a 100Mbit network
By calculation (on a duplex link):
100 Mbit/s = 12.5 Mbytes/s
6300 seconds x 12.5Mbytes/s
-------------------------------------
78,750 Mbytes in 1hr 45mins (maximum)
=====================================
4.5 Gb is less than 6% throughput.
############
On investigating, we found that our Cisco switch is handling loads of
RUNT
ethernet packets:
GLA1SWA> (enable) sh port 4/19
* = Configured MAC Address
Port Name Status Vlan Duplex Speed Type
----- -------------------- ---------- ---------- ------ -----
------------
4/19 [whatever] connected 4 full 100 10/100BaseTX
Port AuxiliaryVlan AuxVlan-Status InlinePowered PowerAllocated
Admin Oper Detected mWatt mA @42V
----- ------------- -------------- ----- ------ -------- ----- --------
4/19 none none auto off no 0 0
Port Security Violation Shutdown-Time Age-Time Max-Addr Trap
IfIndex
----- -------- --------- ------------- -------- -------- --------
-------
4/19 disabled shutdown 0 0 1 disabled
73
Port Num-Addr Secure-Src-Addr Age-Left Last-Src-Addr
Shutdown/Time-Left
----- -------- ----------------- -------- -----------------
------------------
4/19 0 - - - -
-
Port Flooding on Address Limit
----- -------------------------
4/19 Enabled
Port Broadcast-Limit Multicast Unicast Total-Drop Action
-------- --------------- --------- ------- --------------------
------------
4/19 - - - 0
drop-packets
Port Send FlowControl Receive FlowControl RxPause TxPause
admin oper admin oper
----- -------- -------- --------- --------- ---------- ----------
4/19 off off off on 0 0
Port Status Channel Admin Ch
Mode Group Id
----- ---------- -------------------- ----- -----
4/19 connected off 283 0
Port Status ErrDisable Reason Port ErrDisableTimeout Action on
Timeout
---- ---------- ------------------- ----------------------
-----------------
4/19 connected - Enable No Change
Port Align-Err FCS-Err Xmit-Err Rcv-Err UnderSize
----- ---------- ---------- ---------- ---------- ---------
4/19 0 46 0 0 0
############# SEE BELOW #############
Port Single-Col Multi-Coll Late-Coll Excess-Col Carri-Sen Runts
Giants
----- ---------- ---------- ---------- ---------- ---------
-------- ---------
4/19 0 0 0 0 0
22245 0
#####################################
Port Last-Time-Cleared
----- --------------------------
4/19 Wed Feb 8 2006, 12:21:13
Idle Detection
--------------
--
GLA1SWA> (enable) B
I am aware that runts are "<64 bit ethernet packets that are too short
and discarded.
These are usually caused by collisions and may also be caused by poor
wiring and
electrical interference"
The collision rate on the new box is certainly quite high
# netstat -i 1 10
input ce0 output input (Total) output
packets errs packets errs colls packets errs packets errs colls
447302 0 229084 638 21518 447383 0 229165 638 21518
250 0 127 1 13 250 0 127 1 13
512 0 256 0 22 512 0 256 0 22
632 0 352 1 29 632 0 352 1 29
205 0 96 0 11 205 0 96 0 11
588 0 288 1 22 588 0 288 1 22
662 0 352 1 25 662 0 352 1 25
524 0 256 1 23 524 0 256 1 23
785 0 384 1 33 785 0 384 1 33
277 0 160 0 12 277 0 160 0 12
#
The data is transferring to the above box and yet the collisions are on
outgoing packets.
Cisco say:
A collision is the mechanism used by Ethernet to control access and
allocate shared
bandwidth among stations that want to transmit at the same time on a
shared medium.
Because the medium is shared, a mechanism must exist where two stations
can detect
that they want to transmit at the same time. This mechanism is collision
detection.
All that exists between the 100Mbit (configured) vlan switch and the
100Mbit card is a
cat 5 ethernet cable so I can't make sense of the above.
I don't understand why the collisions and runts are happening
##################
My NIC is set up as follows:
# ifconfig -a
lo0: flags=1000849<UP,LOOPBACK,RUNNING,MULTICAST,IPv4> mtu 8232 index 1
inet 127.0.0.1 netmask ff000000
ce0: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
inet [whatever] netmask ffffff00 broadcast [whatever]
ether [whatever]
#
# kstat ce:0 | grep -i link
link_asmpause 0
link_duplex 1
link_pause 0
link_speed 100
link_T4 0
link_up 1
#
#####################
Here are the I/O devices - Both NICs pci108e,abba (network):
================================= IO Devices
=================================
Bus Freq
Brd Type MHz Slot Name Model
--- ---- ---- ---------- ----------------------------
--------------------
0 pci 66 MB pci108e,abba (network) SUNW,pci-ce
0 pci 33 MB isa/su (serial)
0 pci 33 MB isa/su (serial)
0 pci 33 MB isa/rmc-comm-rmc_comm (seria+
0 pci 33 MB pci10b9,5229 (ide)
0 pci 66 MB pci108e,abba (network) SUNW,pci-ce
0 pci 66 MB scsi-pci1000,30 (scsi-2) LSI,1030
0 pci 66 MB scsi-pci1000,30 (scsi-2) LSI,1030
============================ Memory Configuration
============================
######################
Here are the ethernet Drivers and versions that I am using:
# prtconf -PDv | grep -i 'network'
network, instance #0 (driver name: ce)
network, instance #1 (driver name: ce)
#
# modinfo | grep -i ethernet
91 78042000 1a6a6 222 1 ce (CE Ethernet Driver v1.121)
92 1034702f 2bd7 - 1 vlan (Ethernet VLAN module (v1) 1.18)
#
####################
I am using the ce0 interface, though I am aware that ce1 interface may
be
more reliable with v440s (!) from the support forums. However, this is
in the context of a Gigabit ethernet, while I am working with 100Mbit -
i.e.,
http://supportforum.sun.com/hardware/index.php?t=msg&th=4532&start=0&rid
=0
<http://supportforum.sun.com/hardware/index.php?t=msg&th=4532&start=0&ri
d=0>
The sending and receiving ftp hosts are not at all busy.
######################
The best options I can come-up with is
(1) Install latest Solaris 9 NIC driver for my NICs
(2) Switch to ce1 and try that instead.
I would greatly appreciate any information or insights into this problem
that you can give.
I will certainly post any useful info received.
Thanks,
Barry
Visit us at http://www.aggreko.com
Confidentiality Notice: This communication and any accompanying attachments
contain confidential information intended for a specific individual and
purpose. This communication is private and protected by law. If you are not
the intended recipient, you are hereby respectfully notified that any
disclosures, copying, forwarding or distribution, or the taking of any action
based on the contents of this communication is strictly prohibited.
_____________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers
This archive was generated by hypermail 2.1.7 : Wed Apr 09 2008 - 23:38:53 EDT