link based multipathing

From: Unix Administrator (earlysame55@gmail.com)
Date: Wed Mar 12 2008 - 07:26:18 EST


Dear managers,

OS Solaris 10 6/06

Trying to implement link based multipathing.

the 2 config files are:

saturn@root[] cat /etc/hostname.ce1
192.168.144.46 netmask + broadcast + group public up

saturn@root[] cat /etc/hostname.ce5
192.168.144.47 netmask + broadcast + group public up

The host file is:

saturn@root[] cat /etc/hosts
#
# Internet host table
#
127.0.0.1 localhost
192.168.144.47 saturn loghost
192.168.177.58 saturn-bak
#Modification for Link based IPMP
192.168.144.46 saturn-ce1

ifconfig output:

saturn@root[] ifconfig ce1
ce1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 2
        inet 192.168.144.46 netmask ffffff00 broadcast 192.168.144.255
        groupname public
        ether 0:4:ab:ac:bd:16

saturn@root[] ifconfig ce5
ce5: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
        inet 192.168.144.47 netmask ffffff00 broadcast 192.168.144.255
        groupname public
        ether 0:5:ab:cf:8a:16

PROM setting
local-mac-address?=true

now the problem, when a failed over is initiated from the command line:

saturn@root[] ifconfig ce1
ce1: flags=89000842<BROADCAST,RUNNING,MULTICAST,IPv4,NOFAILOVER,OFFLINE> mtu
0 index 2
        inet 0.0.0.0 netmask 0
        groupname public
        ether 0:4:ab:ac:bd:16
saturn@root[] ifconfig ce5
ce5: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
        inet 192.168.144.47 netmask ffffff00 broadcast 10.20.144.255
        groupname public
        ether 0:5:ab:cf:8a:16

saturn@root[] ifconfig ce5:1
ce5:1: flags=1000843<UP,BROADCAST,RUNNING,MULTICAST,IPv4> mtu 1500 index 4
        inet 192.168.144.46 netmask ffffff00 broadcast 10.20.144.255

With the above configuration when i fallback to the original interface i get
a timeout for a shortwhile, which is about less than 8 secs. This does not
happen in probe based. The test was done from the switchside where we
brought down the ports down and then brought it up. This connection loss was
on the continous ping and telnet session exiting.
The other question is it appears that the interfaces are not load balanced,
netstat -i reports activity on both interfaces but if a snoop is done it
does not show results in the secondry interface.

output of

snoop -d ce1 -V 192.12.43.72
Using device /dev/ce (promiscuous mode)

[ nothing ]

snoop -d ce5 -V 192.12.43.72

but when i do

_______________________________
saturn.hq.martin.sample.com -> PC.hq.martin.sample.com ETHER Type=0800 (IP),
size = 54 bytes
saturn.hq.martin.sample.com -> PC.hq.martin.sample.com IP D=192.12.43.72 S=
192.168.144.47 LEN=40, ID=45733, TOS=0x0, TTL=60
saturn.hq.martin.sample.com -> PC.hq.martin.sample.com TCP D=3085 S=6515
Ack=1310135297 Seq=2637315031 Len=0 Win=49880

Please help me to figure out what's going on.

Thanks for all the help.
_______________________________________________
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 - 23:42:50 EDT