"Graham" == Graham C Knight grahamk@ast.lmco.com writes:

Graham> We've run into some trouble with the onboard FC-AL on Graham> F700's. It's interesting that they are now shipping with Graham> PCI cards installed.

Graham> Will Netapp be sending people free PCI FC-AL cards for Graham> existing F700 systems?

NA just sent me a pair of PCI FC-AL's to replace the on-board FC-AL's. Apparently, the on-board FC-AL in filer A detected a loop break on its partners disks and that caused filer A to failover to filer B. It seems odd that filer A would be the one to failover upon detecting a loop break on its secondary FC-AL loop.

I've got a question about upgrading the FC-AL controllers though.

Two options for upgrading:

1) turn everything off, install the FC-AL contollers, move each filer's A-loop, turn everthing back on.

2) failover filer A, turn filer A off, upgrade its FC-AL controller, move its A loop, turn it back on, giveback to filer A from filer B. Then do the same on filer B.

This should be no different then adding a shelf to a cluster. Last time we did that, we used scenario (2). It didn't go smoothly, but that's because we were bitten by the 44+ days uptime bug.

j.

NetApp support informed me that they now ship all F760's with the onboard FC-AL adapter terminated and add an extra pci FC-AL adapter instead. I was also told that the onboard controller most likely caused several problems that we experienced like: 'watchdog resets', 'UNCORR PROC 98' errors, Netapp momentarily disappearing from the network, and 'disk underrun' errors. Terminating the onboard controller and adding a new pci FC-AL controller has seemed to fix all these problems for us.

David Midgett Manager, Server Operations About.com, Inc. The Leading Network of Niche Vertical Sites. http://About.com

-----Original Message----- From: owner-toasters@mathworks.com [mailto:owner-toasters@mathworks.com]On Behalf Of Brian Tao Sent: Thursday, November 11, 1999 7:22 PM To: bentele@mtu-friedrichshafen.com Cc: toasters@mathworks.com; jay@cimedia.com Subject: Re: AW: 'Loop break detected' followed by failover

On Wed, 10 Nov 1999 bentele@mtu-friedrichshafen.com wrote:

we also had a ' isp2100_timeout]: Resetting ISP2100 in slot ..' a few weeks ago. F740 , DOT 5.2.3, FC-AL 18 GB, Gbit Card.

!! Be careful !!

Replacing motherboard and FC-Adapter-Card at last solved the problem, after we replaced all cables, all shelves of this loop, and about 3 disks.

I think there is a known (but perhaps not widely acknowledged) problem with the on-board FC-AL interface on the F740 motherboards (and possibly on other models as well). I have four F740's in production, of which two have had histories of flaky FC-AL (ISP2100 timeouts during disk scrubs, hung RAID reconstructions, simultaneous errors across all drives on a shelf, etc.). Netapp has sent up four slot-based FC-AL adapters and told me to use those instead of the on-board ones.

Another shipment of four F740's arrived earlier this week, and I noticed they came with two slot-based FC-AL adapters each (these are clustered systems). I take that as an indication that Netapp still believes there is some sort of defect on the F740 motherboards. -- Brian Tao (BT300, taob@risc.org) "Though this be madness, yet there is method in't"