FC100/S - a piece of crap?

Ok, that might be a bit to harsh, but hey - I'm annoyed right now :-)

Situation:

Sun Enterprise 3500 with two of the old I/O boards (with only the 25Mbps
FC links onboard) with two FC100/S boards (501-5266) attached to
two Sun A5000 arrays filled with 36GB Seagate Cheetah disks.

*Everything* has the latest firmware. Solaris 9 with all the latest
patches too.

Now, whenever we put any kind of load on the FC links to the arrays
we start seeing a lot of FC and disk not responding errors. So many
that it's practically useless for the intended purpose (HOME server
for the local university computer club who got the servers and
arrays from a local company).

Doesn't matter if I only use one of the FC100/S boards. Doesn't matter
which I/O motherboard we use. We swapped all the GBICs. Even tried
with a pair of brand new copper GBICs. Still no go.

So I figured it might be something with the E3000 server itself and so I
took an Ultra60 running Solaris 9 with all the latest patches and installed
an Emulex FC HBA adapter into it. Worked like a charm. Not a single error/failure.

Ok. Next try - grabbed an Ultra2/2x300MHz/1024MB and installed a fresh
Solaris 9 into it. Applied all available patches. Took one of the FC100/S
boards and installed into it. Booted. The disks show up just fine, enter "format",
select one of the disk and then select "diag" and "read". Then they fail
after a very short while with the following errors in the syslog:

>Jun 24 14:14:08 inservitus.lysator.liu.se socal: [ID 403145 kern.warning] WARNING: ID[SUNWssa.socal.link.6040] socal0: SOCAL:Forcing SOC+ reset as LIP timed out
>Jun 24 14:14:08 inservitus.lysator.liu.se socal: [ID 403145 kern.info] ID[SUNWssa.socal.driver.1010] socal0: Downloading host adapter, fw date code: @(#)Thu May 3 18:50:33 2001
>Jun 24 14:14:09 inservitus.lysator.liu.se socal: [ID 403145 kern.info] ID[SUNWssa.socal.link.6010] socal0: port 1: Fibre Channel Loop is ONLINE
>Jun 24 14:14:09 inservitus.lysator.liu.se socal: [ID 403145 kern.info] ID[SUNWssa.socal.link.5010] socal0: port 1: Fibre Channel is OFFLINE
>Jun 24 14:14:09 inservitus.lysator.liu.se socal: [ID 403145 kern.info] ID[SUNWssa.socal.link.6010] socal0: port 1: Fibre Channel Loop is ONLINE
>Jun 24 14:14:20 inservitus.lysator.liu.se scsi: [ID 243001 kern.info] /sbus@1f,0/SUNW,socal@3,0/sf@1,0 (sf1):
>Jun 24 14:14:20 inservitus.lysator.liu.se REPORT_LUNS to target 0x1d retrying

(Also tried swapping GBICs in this setup - tried both the copper GBIC and the
Sun original fiber GBICs, no difference at all).

I spent some time just now and searched Sunsolve and Google - and it seems there
is a fair amount of people having problems with the FC100/S boards and the A5000.

Some things that helped some people was using the modern Ex000 I/O board with the
built-in SOC+ interface and just forget about the FC100/S boards...

Has anyone else seen these errors and have been able to solve them?

- Peter



--
--
Peter Eriksson <> Phone: +46 13 28 2786
Computer Systems Manager/BOFH Cell/GSM: +46 705 18 2786
Physics Department, Linköping University Room: Building F, F203