|
Home > Archive > Data Storage > November 2006 > Brocade / Cisco interop issue
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
Brocade / Cisco interop issue
|
|
| Jon Metzger 2006-10-25, 7:12 pm |
|
Hey folks. This might be a long shot but I'm hoping there might be a
Brocade guru out there that has some idea about the problem I'm seeing.
We've got an all cisco MDS Fibre Channel fabric and someone has
purchased a Dell blade server with 2 Brocade 3014 14-port switches in
it. We created a VSAN on the cisco side in interop mode 1. We modified
buffer credits on the cisco side and the ISL mode on the Brocade side
and the E port comes up fine. So far so good.
In order for the blade server to use the same storage ports on our SAN
as other hosts, we needed to create IVR zones. This is pretty
straightforward and it worked as expected -- hosts logged in and saw
storage. What's strange is that whenever an individual blade reboots,
the Brocade switch will not allow it to bring its FC port up. A
"showport" on the Brocade says "Switch not ready for F or L ports". If
I drop the ISL between the cisco and the Brocade, the Brocade lets me
"portenable" the blades again. I can then bring the ISL back up and the
blade servers can get to their storage again.
It's not reasonable to have to drop an ISL every time one of ten blades
reboots, so I need to figure this out. I've had one support channel
punt on the problem already, and I'm only familiar with running Cisco
Fibre Channel gear. Any Brocade-ites out there have an idea about what
could be wrong if the switch reports "Switch not ready for F or L ports"?
A few details:
Cisco MDS 9509, firmware 2.1.2b
-Enterprise license
Brocade 3014, firmware 4.4.1b
-Full fabric license
| |
| Jon Metzger 2006-10-26, 1:12 pm |
| Jon Metzger wrote:
>
> Hey folks. This might be a long shot but I'm hoping there might be a
> Brocade guru out there that has some idea about the problem I'm seeing.
>
> We've got an all cisco MDS Fibre Channel fabric and someone has
> purchased a Dell blade server with 2 Brocade 3014 14-port switches in
> it. We created a VSAN on the cisco side in interop mode 1. We modified
> buffer credits on the cisco side and the ISL mode on the Brocade side
> and the E port comes up fine. So far so good.
>
> In order for the blade server to use the same storage ports on our SAN
> as other hosts, we needed to create IVR zones. This is pretty
> straightforward and it worked as expected -- hosts logged in and saw
> storage. What's strange is that whenever an individual blade reboots,
> the Brocade switch will not allow it to bring its FC port up. A
> "showport" on the Brocade says "Switch not ready for F or L ports". If
> I drop the ISL between the cisco and the Brocade, the Brocade lets me
> "portenable" the blades again. I can then bring the ISL back up and the
> blade servers can get to their storage again.
>
> It's not reasonable to have to drop an ISL every time one of ten blades
> reboots, so I need to figure this out. I've had one support channel
> punt on the problem already, and I'm only familiar with running Cisco
> Fibre Channel gear. Any Brocade-ites out there have an idea about what
> could be wrong if the switch reports "Switch not ready for F or L ports"?
>
> A few details:
>
> cisco MDS 9509, firmware 2.1.2b
> -Enterprise license
> Brocade 3014, firmware 4.4.1b
> -Full fabric license
An update to my own problem... I took IVR out of the mix to see if the
problem was IVR related and I ran into another problem which I think is
likely related.
My new configuration is one Brocade 3014 with one host attached and one
Cisco MDS 9509 with one host attached. Both switches are in interop
mode. A fresh reboot is done on the Brocade switch and everything shows
up and works correctly. On the Brocade, running "nsallshow" I see the
FCIDs of both hosts. If I disable and then re-enable the host port on
the cisco switch, the output of "nsallshow" and "nsshow" on the Brocade
return nothing. Within 15 seconds or so of cycling the host port on the
Cisco switch the Brocade switch *panics* and reboots itself.
I'm going to go back to our vendor support with this new information and
see if they will listen to me now. Any ideas anyone?
| |
| Moojit 2006-10-27, 1:13 am |
| By bouncing the host port connected to the cisco switch you're causing the
local and global directory name servers on both switches to remove/add
objects.
Objects are removed when the node goes away, objects are added when the node
logs back into the fabric. The cisco switch will send these NS updates to
Brocade switch via it's ISL link to update
it's NS database. Try adding some time between the remove / add of the host
to see if this removes the Brocade failure. If you're doing this quickly,
you may be bringing out an issue on the Brocade
switch.
Da Moojit
"Jon Metzger" <blagh@___gmail.com> wrote in message
news:ehqmqu$3r1$1@lenny.tc.umn.edu...
> Jon Metzger wrote:
>
> An update to my own problem... I took IVR out of the mix to see if the
> problem was IVR related and I ran into another problem which I think is
> likely related.
>
> My new configuration is one Brocade 3014 with one host attached and one
> cisco MDS 9509 with one host attached. Both switches are in interop mode.
> A fresh reboot is done on the Brocade switch and everything shows up and
> works correctly. On the Brocade, running "nsallshow" I see the FCIDs of
> both hosts. If I disable and then re-enable the host port on the Cisco
> switch, the output of "nsallshow" and "nsshow" on the Brocade return
> nothing. Within 15 seconds or so of cycling the host port on the Cisco
> switch the Brocade switch *panics* and reboots itself.
>
> I'm going to go back to our vendor support with this new information and
> see if they will listen to me now. Any ideas anyone?
| |
| Jomait 2006-11-01, 1:13 pm |
| Hi,
I am a reporter with www.searchstorage.com and was interested to read
that you guys were experiecing problems with switches from cisco and
Brocade interoperating in the same fabric. I thought that "interop
mode" fixed these issues, but it doesn't sound like?
If anyone is willing to follow-up, I'd be glad to pursue a story on
this.
Cheers,
Jo
Moojit wrote:[vbcol=seagreen]
> By bouncing the host port connected to the cisco switch you're causing the
> local and global directory name servers on both switches to remove/add
> objects.
> Objects are removed when the node goes away, objects are added when the node
> logs back into the fabric. The cisco switch will send these NS updates to
> Brocade switch via it's ISL link to update
> it's NS database. Try adding some time between the remove / add of the host
> to see if this removes the Brocade failure. If you're doing this quickly,
> you may be bringing out an issue on the Brocade
> switch.
>
> Da Moojit
> "Jon Metzger" <blagh@___gmail.com> wrote in message
> news:ehqmqu$3r1$1@lenny.tc.umn.edu...
|
|
|
|
|