Summary: hba and attached device management

Due to some political issues, I've not been able to test and confirm
details wrt responders suggestions/questions.  If the climate relaxes
and system access becomes available, I'll post some test results for
my situation.  Until that time comes, I'll post this preliminary summary
of in case there are others are interested.  Sorry, I can't follow
through now.

For clarity, the "failed" state I'm refering to is in the "Condition"
column of output from the cfgadm -al.  The following output is not from
the hurt system, but rather from a system I have access to at this time,
and I typed failed in here, just to make the point.  I can't remember
the values in the other columns, but I'm sure "connected" was not
in the mix:

# cfgadm -al
Ap_Id                          Type         Receptacle   Occupant     Condition
c0                             scsi-bus     connected    configured   unknown
c0::dsk/c0t2d0                 disk         connected    configured   unknown
c0::dsk/c0t3d0                 disk         connected    configured   unknown
c4                             fc           connected    unconfigured failed     <- badness happens here
c5                             fc           connected    unconfigured unknown


Now for the kind, first responders:

1. Vikash Gupta suggested that I suppress RSCN's on zone set activations.
My cisco (switch manufacturer) said, "Although I've never heard of
using the RSCN suppression command to alleviate issues around zoning,
I can't say that it would specifically hurt anything."

2. Jason Murray pointed to this web page: 
     http://unixhosts.us/txt/attached-solaris-to-san-without-rebooting.txt
which lists the cfgadm commands that we had been using without
success in recovering the t2000 hba.  Jason said he does this
activity all the time on his Xiotech SAN and t2000 without hba problems.    
This is good to know.  It means we should be able to change something
and get where we need to be.

3. Johan Hartzenberg wrote some "untested" comments that I'd love to get
back in the systems to try.  Johan said, "I think your problem is switch
model and even switch firmware specific."  I can't even get this information
right now, so I'm kind of stuck, but this idea might be consistant
with Jason's response.  He said that the hba "might" get recovered with:

  cfgadm -c deconfigure c2
  cfgadm -c configure c2

and possibly:

  cfgadm -c disconnect c2
  cfgadm -c connect c2

Another idea from Johan would be to use modunload the driver before making
zoning changes, and then after making your changes, do a modload.
_______________________________________________
sunmanagers mailing list
sunmanagers@sunmanagers.org
http://www.sunmanagers.org/mailman/listinfo/sunmanagers