Problem getting SUN trunking to work

Hi all,

I am trying to setup SUN trunking per the instructions in the SUN Trunking 1.3 Installation and User's Guide. Dated June - 2007.

I am setting it up in a test environment. Pretty simple test bed. 2 Sun V490s connected to a Catalyst 6500 via fiber NICs. 2 NIC's in each 490. Each is a Gigaswift (ce) type NIC.

Patch levels on the boxes are as follows.

SunOS server01 5.10 Generic_127127-11 sun4u sparc SUNW,Sun-Fire-V490
SunOS server02 5.10 Generic_127127-11 sun4u sparc SUNW,Sun-Fire-V490

In the Guide it states, " Sun Trunking 1.3 ... provides the ability to aggregate multiple links so that they work in parallel as if they were a single link."

That is what I want to do. So... following along in the guide....

I have checked that both required packages are present.

server01/>#pkginfo -l SUNWtrku
PKGINST: SUNWtrku
NAME: Sun Trunking Utility 1.3 for Solaris 10
CATEGORY: application
ARCH: sparc
VERSION: 1.3,REV=2005.06.14.5.10
BASEDIR: /
VENDOR: Sun Microsystems, Inc.
DESC: Sun Trunking PCI/SBus Adapter Utility
PSTAMP: miro20050614090857
INSTDATE: Feb 27 2008 15:56
HOTLINE: Please contact your local service provider
STATUS: completely installed
FILES: 25 installed pathnames
11 shared pathnames
16 directories
5 executables
572 blocks used (approx)

server01/>#pkginfo -l SUNWtrkm
PKGINST: SUNWtrkm
NAME: Sun Trunking Utiltiy 1.3 Man Pages
CATEGORY: system
ARCH: all
VERSION: 1.3,REV=2005.06.14.10.0
BASEDIR: /opt
VENDOR: Sun Microsystems, Inc.
DESC: Sun Trunking PCI/SBus Adapter man pages
PSTAMP: miro20050614090918
INSTDATE: Feb 27 2008 15:56
HOTLINE: Please contact your local service provider
STATUS: completely installed
FILES: 12 installed pathnames
8 directories
44 blocks used (approx)

server01/>#

I found my instance numbers....

server01/>#grep net /etc/path_to_inst
"/pci@8,700000/network@4" 0 "ce"
"/pci@8,700000/network@5" 3 "ce"
"/pci@9,700000/network@2" 1 "ce"
"/pci@9,600000/network@1" 2 "ce"
server01/>#

OR

server01/>#dladm show-dev
ce0 link: unknown speed: 1000 Mbps duplex: full
ce3 link: unknown speed: 1000 Mbps duplex: full
ce1 link: unknown speed: 1000 Mbps duplex: full
ce2 link: unknown speed: 1000 Mbps duplex: full
server01/>#

I'm going to use ce0 and ce3 for the trunk.

Added the following line the the nettr.sh script

nettr -setup 944 device=ce members=0,3 policy=1 lacp=p ptimer=1

I created the appropriate hostname file in /etc/

/etc/hostname.ce0

As I said, the switch that these 2 boxes are connected to is a Cisco Catalyst 6500. Config is as follows....

SW-Cat6506-A> (enable) sh port lacp-channel info
Switch Frame Distribution Method: ip both

Port Status Channel Admin Channel Speed Duplex Vlan
mode group id
----- ---------- -------------------- ----- ------- ----- ------ ----
4/6 connected active 944 1764 1000 full 150
4/8 connected active 944 1764 1000 full 150

Port Channel Oper-group Neighbor Oper-Distribution PortSecurity/
ifIndex Oper-group Method Dynamic port
----- ------- ---------- ---------- ----------------- -------------
4/6 177 15105 945 ip both
4/8 177 15105 945 ip both

Port Device-ID Port-ID Platform
----- ------------------------------- ------------------------- ----------------
4/6 00-13-bc-31-b9-fa 51203
4/8 00-13-bc-31-b9-fa 51200

So everything appears to be setup per the instructions in the guide, however, when I boot the box and it tries to establish the aggregate link, I see the following error message.

NOTICE: trunk link (ce3): Partner Key=0: Please configure a valid key value on switch.

NOTICE: trunk link (ce0): Partner Key=0: Please configure a valid key value on switch.

NOTICE: trunk link (ce3): Partner Key=0: Please configure a valid key value on switch.

NOTICE: trunk link (ce0): Partner Key=0: Please configure a valid key value on switch.

NOTICE: trunk link (ce3): Port Partner MAC (0:7:d:72:1e:0) or key (15106) incompatible with Aggregation Partner MAC (0:7:d:72:1e:0) or key (15105)

My question is, why is it saying my Partner Key=0 when I see in the switch config that it is 944. What am I missing? I did confirm that the MAC address (0:7:d:72:1e:0) is in fact, the MAC address of the switch port where this box is plugged in. Not sure where the other keys 15105 and 15105 are coming from.

Any help on this would GREATLY be appreciated.
Thanks
Bruhn

Unless a V490 is a Java networking class or API you're asking in the wrong place.

Don't understand your post dude. The title of this forum is The UNIX and Linux Forums -> OS Specific Forums -> SUN Solaris.

SUN Trunking = SUN Solaris = OS Specific. Besides, I see other posts in here about VMware, Networking, etc.... and you replied to them.

Sure you're in the right place?

Did you try configuring using a PAgP protocol instead? Can have a look at you script? And alos perform a tcpdump to see if the packeys are sent/recieved.

In any case the trunking commands are applied to the individual interface and not to the channel.

1.Take out the trunk commands.

2.Read about PaGP, figure out if SUN wants to (or indeed can) negotiate EtherChannel.

You could try setting the switch EtherChannel to ON.

post your output for the following commands

ifconfig -a
dladm show-aggr -L
dladm show-aggr -s

as for the 6506 switch... i can't help you with some output as i have a 6509. syntax and output are different. (sh lacp neighbor detail)

incredible... will pagp work with solaris ? i've only known (just from experience) lacp as teh aggr-link protocol. can the switch use pagp and solaris use lacp?

I have not tried that, but I suppose it could be done.

HI,

read Document ID: 6315863 @ sunsolve

Work Around:

During evaluation of this bug it has been found that you savely may ignore this messages, which happen during initialization.

Could be... As it is showing 'NOTICE' and not either ERROR or WARNING ?

Hey thanks for all your help folks.

I spoke with SUN and they confirmed that this is in fact a bug. Taran posted the bug ID above. They said to ignore the message.

Also, pagp will not work. Solaris doesn't support it. Both the switch and the host must use LACP.

Thanks again for all you help.
Bruhn