Can't ping external IP from two interfaces on SLES 11 SP2

I’m trying to set up multi-path I/O connection between a SLES 11 SP2 server and a Dell EqualLogic SAN. Dell has a White paper called “Deploying SUSE Linux Enterprise Server with Dell EqualLogic PS Series Storage Arrays” and the first step is to bind unique IP addresses to two NICs and confirm that you can ping the EqualLogic SAN. I have a Dell PowerEdge R720 server with 4 NICs. The first NIC goes to the production network and NICs 3 and 4 go to the iSCSI network and Dell switches then to the EquaLlogic box. Whichever of the two NIC’s I bind first can ping the EqualLogic without a problem. The NIC that is bound second can initially not ping the EqualLogic at all. After about 10 to 15 minutes of the srever being up, it can ping the EqualLogic as well.

I can use ifdown to shut down NIC 3 and NIC 4 then ifup to start it up. If I shut down NIC 3 and then NIC 4 then bring up NIC 4 first then NIC 3, now NIC 4 can ping the box and NIC 3 can’t.

Dell whitepaper says to not proceed with loading their Multipath IO so I haven’t attempted to load that yet.

I can’t see any reason for this behavior. Does anyone have an idea why the first NIC bound will work but the second NIC takes over 10 minutes to respond to a ping?

Dan

dchunt wrote:
[color=blue]

I can’t see any reason for this behavior.[/color]

Can you provide additional information about your configuration
including IP addresses?

Have you checked the Knowledgebase?
https://www.suse.com/support/

Depending on your configuration, this may help but I’m only guessing
until I know more.

TID 7000318 Reply packets are sent over an unexpected interface
http://www.suse.com/support/kb/doc.php?id=7000318


Kevin Boyle - Knowledge Partner
If you find this post helpful and are logged into the web interface,
show your appreciation and click on the star below…

Kevin, I’m not sure exactly what you need. Our SLES server has 4 NIC’s in it - em1 through em4. em1 is connected to the production network -172.16.x.x. em2 is not used. em3 and em4 are both connected to the iSCSI network on the same subnet as the EqualLogic. em3 is 192.168.x.y and em4 is 192.168.x.z. By default during boot, em3 initializes before em4. Therefore if I try to ping the EqualLogic through em3 using:

ping -I em3 192.168.x.w

I get a response immediately. If I then do the same ping but instead ping out em4 I don’t get anything for the first 10 to 15 minutes the server is up, then it will start pinging.

Now, if I do an ifdown for both em3 and em4, then do ifup em4 and then try to ping, I get a response immediately from the EqualLogic. If I then bring up em3 and try to ping, now it won’t get responses. So the second interface that comes up is the one having the problem - it is not the physical NIC itself. BTW, I tried using em2 in place of em4 and I get the same results.

According to the white paper mentioned in the first post, I should be able to ping from both NICs to the EqualLogic and this should be working before I install the Dell multi-IO. Do you agree that I should be able to ping from both NICs and get a response?

Thanks,

Dan

dchunt wrote:
[color=blue]

According to the white paper mentioned in the first post, I should be
able to ping from both NICs to the EqualLogic and this should be
working before I install the Dell multi-IO. Do you agree that I
should be able to ping from both NICs and get a response?[/color]

Yes! I don’t know why it isn’t working. :frowning:

To find out what’s going on, I’d run WireShark and get a trace so I’d
know for sure. Other than that you could try a few things and see if
they help.

The first place to look is the Knowledgebase:
https://www.suse.com/support/

A search for “interface” returns an assortment of TIDs. You may spot
something there.

Here is one in particular that is interesting:
TID 7007649 Applying SLES 11 SP 1 Causing Communication Issues
http://www.novell.com/support/kb/doc.php?id=7007649

Let me know if it helps or if you if uncover any additional clues.


Kevin Boyle - Knowledge Partner
If you find this post helpful and are logged into the web interface,
show your appreciation and click on the star below…

Thanks for the reply Kevin. This is a brand new SLES 11 SP2 build. Right now 3 Dell engineers are working on it and scratching their heads. They brought their own OS (probably Windows or Redhat) and are getting the same results. I’ll let you know if I find out what the issue is.

Dan

The problem was that you HAD to load the Dell multi-io software before SLES would ping the EqualLogic from either interface. This was exactly the opposite of what the Dell Whitepaper said that you had to do.

Dan

dchunt wrote:
[color=blue]

The problem was that you HAD to load the Dell multi-io software before
SLES would ping the EqualLogic from either interface. This was
exactly the opposite of what the Dell Whitepaper said that you had to
do.

Dan[/color]

That’s good to know. I’m looking at the EqualLogic for a customer and
trying to do my homework so I don’t get caught in a similar situation.
:frowning:

I’m glad you got it working. Still, you might want to search the
knowledgebase as there have been some issues involving multi-path. It
never hurts to be aware of them just in case…


Kevin Boyle - Knowledge Partner
If you find this post helpful and are logged into the web interface,
show your appreciation and click on the star below…