[f-nsp] trunk problems on a HP9304 / Foundry BigIron4000
Asoka De Saram
adesaram at yahoo.com
Wed Oct 17 11:02:11 EDT 2007
Ramon,
If you define the trunk as a server trunk the load balancing algorithm will take TCP/UDP source & destination ports in to consideration. So you will get a better distribution of traffic by recreating trunk as a server trunk.
The syntax:
trunk server ethe 1/1 to 1/2
Sent via BlackBerry from T-Mobile
-----Original Message-----
From: Ramon van Alteren <ramon at hyves.nl>
Date: Wed, 17 Oct 2007 16:53:07
To:foundry-nsp at puck.nether.net
Cc:"'frank'" <frank at hyves.nl>
Subject: [f-nsp] trunk problems on a HP9304 / Foundry BigIron4000
Hello List,
We employ 4 HP9304M routers which are in fact relicensed foundry bigiron
4000 switches, and we are having problems with defining trunks on one of
the routers.
I am hoping that the problem sounds familiar to someone on the list.
HP firmware appears to follow foundry firmware versions, current version
is: SW: Version 08.0.01cT53 Hewlett-Packard Company
We have defined a similar trunk on two routers, the routers are near
identical copy wrt configuration. They function as primary/secondary
routers in our infrastructure.
On both routers the trunk is created as follows:
SSH at router4#sh trunk e 1/1 to 1/2
Max number of server trunks: 32; available: 32
Number of hash buckets per server trunk: 256
Configured number of hash buckets per server trunk: 256
Configured trunks:
Trunk ID: 1
Type: Switch
Ports_Configured: 2
Primary Port Monitored: Jointly
Ports 1/1 1/2
Port Names uplink* uplink*
Port_Status enable enable
Monitor off off
Mirror Port N/A N/A
Monitor Dir N/A N/A
The original command was: trunk switch ethernet 1/1 to 1/2
On the primary router the traffic is balanced correctly over both fiber
links. On the secondary router the traffic is send over one of the two
fibers in the trunk.
If I bring the interface of the fiber which carries the traffic down on
the router the traffic is rerouted to the other interface in the trunk.
I can find no commands that influence traffic balancing for layer2
switch trunks in the HP or Foundry docs and we are at loss how to
resolve the situation.
So far we've tried:
* removing and recreating the trunks
* reloading the router
Neither have solved the problem.
Since we have outgoing traffic in excess of 1.4gbit/s we'd like to make
sure that our secondary router starts balancing the traffic similar to
our primary router.
We suspect that the primary router balances based on layer3 and the
secondary router doesn't for some reason. This suspicion is based on the
output of the show trunk command:
primary router:
Operational trunks:
Trunk ID: 1
Type: Switch
Duplex: Full
Speed: 1G
Tag: No
Priority: level0
Active Ports: 2
Ports 1/1 1/2
Link_Status active active
port_state Forward Forward
Load Sharing
Mac Address 2 2
IP 44295 44294
IPX 0 0
Apple Talk 0 0
Multicast 0 0
secondary router:
Operational trunks:
Trunk ID: 1
Type: Switch
Duplex: Full
Speed: 1G
Tag: No
Priority: level0
Active Ports: 2
Ports 1/1 1/2
Link_Status active active
port_state Forward Forward
Load Sharing
Mac Address 1 1
IP 0 0
IPX 0 0
Apple Talk 0 0
Multicast 0 0
Neither interfaces have a ip-address defined, both trunks are member of
a single vlan which has a router-interface associated with it. On these
particular interfaces the vlan is sent untagged, but we see similar
behaviour if we sent tagged traffic.
We run vrrp-extended on top of the ve to communicate with upstream and
flap traffic in case of failure.
The interface module in the router is a J4885A EP 8 port mini-GBIC
Management Module and afaik identical to a foundry redundant module.
Any pointers would be most appreciated.
Best Regards,
Ramon van Alteren
Senior System Administrator Hyves.nl
_______________________________________________
foundry-nsp mailing list
foundry-nsp at puck.nether.net
http://puck.nether.net/mailman/listinfo/foundry-nsp
More information about the foundry-nsp
mailing list