I have two catalyst 3560 switches. they are connected via a single trunk port using 082.11q on gi0/23 on each switch.
I'm trying to set up an internal switch infrastructure, such that each of my linux servers are connected to both switches, using etherchannel on the switch, and bonding on the linux host.
For the most part I have this working. say I have "server x" with eth1 and eth0, connected to gi0/1 on both switches. I've configured each port on each switch using "channel-group 1 mode on" and made sure port 1 on each switch is on the same vlan and that the resulting po1 is on the right vlan.
Things usually work fine. I can physically disconnect one interface on the server and still have connectivity. However, I am having intermittent connectivity issues on all the servers.
My switch logs show these for all of my poX groups for my servers on an intermittent basis:
Host (mac srubbed) in vlan 1 is flapping between port Gi0/23 and port Po10
Keep in mind that gi0/23 is my trunk port between the two switches. I don't really understand what the problem is because the logical port info for each poX seems to be passing over the trunk interface and things seem to work most of the time. I've looked this up all over the web and I've tried a few different things:
1) I've tried configured gi0/23 on each switch with "spanning-tree bpdufilter enable". Toggling this on and off doesn't seem to help any.
2) I've tried configuring gi0/23 on each switch with "l2protocol-tunnel point-to-point" and this doesn't seem to help either.
Each time, I can re-create the flapping log entry just be sending traffic to any one of those servers and I'll get that error. The odd thing is that most of the time, the traffic still makes it to and from the host. Only about 20% of time do connections actually get dropped.
I'm using the default bonding method (round-robin) on each host, but I've also tried configuring one of the hosts to use active/standby mode for the bonding option and that didn't seem to help either.
Any ideas?
Running config for both switches:
This Question has been solved and asker verified All Experts Exchange premium technology solutions are available to subscription members.
Experts Exchange has been collecting answers to technology questions since 1996…3 million and counting! If you have a question, chances are we already have your answer.
If you can't find the exact answer you're looking for, ask our exclusive community of 50,000 experts. You’ll get a personalized answer from a trusted professional.
Thousands of free tech tips, tricks, how-to’s and tutorials are available in our peer reviewed articles section. See for yourself how smart our experts are, no login required.
Save solutions to your questions, answers you’ve discovered through searching plus helpful articles in your personal knowledgebase for easy future access.
Access the answers to your technology questions today.
30-day free trial. Register in 60 seconds.
Members of the expert community talk about why the experience at Experts Exchange is different than what you will find anywhere else.
Try it out and discover for yourself.
30-day free trial. Register in 60 seconds.
Join the community of experts here and help other tech pros by answering question in your area of expertise. You can earn FREE access to all Experts Exchange's premium features and resources.
I'm trying to have a redundant switching architecture so our small, simple network can survive a switch being down. The idea is that all hosts are connected to two physical switches, so that if one switch dies, all hosts can still communicate with each other over the other switch.
I'm confused as to why this can't be done. Is it because these switches aren't "stacked" together as one switch? I've set this up before with higher-end catalysts, the only difference was that they were all set up as a single switch stack, so all of the ports were considered to belong to one master, logical switch.
Is that why I can't do this? Is there any way I can have these lower-end 3560's act as a single switch? I see that the 3560's support switch "clustering", but it doesn't seem the same thing as stacking. Could I use clustering as a way to pull this off and pass the layer 2 information across both switches?
Shouldn't STP be stopping one of these links and therefore preventing a loop? Although I guess the server has no idea what STP packets are and is ignoring them so the switch has no idea this is happening.
However from my understanding, EtherChannel was designed to do the following:
switch1 ------------- switch2
\ \
\ \
eth0-eth1
You should also be able to have another two links to switch 2. However this would require more ethernet cards, STP shouldn't matter because your server should never forward packets received on etherchannel 1 out etherchannel 2. I can't say I've ever done it. Normally you create loops between access and distribution or distribution and core layers for redundency. I've never seen a server so critical that it needs to become part of a loop. If there is such a server, it may be time to look at another solution like clustering as the redundent layer 2 approach only works until your server becomes the point of failure and that seams like a lot of switchports and network cards to waste on something that will end up failing cause the fan siezed up.
KevJB:
Not sure what you are trying to say. STP is enabled, but as far as I understood and like you say, STP is ignored from the servers perspective.
As far as the reasons why I'd like to have redundant core switches, is pretty much the same reason why you have redundant anything. A cluster of servers is great when you have a server failure. But if they are both connected to the same physical switch, then a cluster of servers doesn't help much if you have a switch failure.
At any rate, I appreciate the advice as far as topology is concerned, but that really isn't my problem. My problem lies in capability. Like I said, I know that what I'm trying to do works just fine with switch stacks. I've done it before and it worked quite well. We could lose any physical switch in the "stack" and not have any network outages. One of the other nice things about this setup is that it takes a lot of the stress out of doing switch maintenance/upgrades knowing that you can reboot/reconfigure/test one switch without any downtime.
The idea behind the tunneling is to allow you to create an etherchannel "through" a switch to the channeling device at the other end.
An example can be found in figure 15-6
http://www.cisco.com/e
Wh
Your best bet would be to remove the etherchannel commands to the servers and let spanning tree deal with any loops.
Business Accounts
Answer for Membership
by: donjohnstonPosted on 2008-09-30 at 13:14:55ID: 22609026
I'm not quite clear on the topology.
Do you have one server with two network interfaces and each one connected to a different switch?
Switch 1-------------Switch2
\ /
\ /
---- Server-----