From: Jean Marie Diaz (jdiaz@westcoastbeauty.com)
Date: Thu Sep 25 2003 - 14:08:32 EDT
Executive summary: upgrade.
Thanks to Jim Lola, Dr. Thomas Blinn, Alan Rollow, Mike Heiser, and Nikola
Milutinovic for rapid input. Alan's comments were most succinct and I
reproduce them below:
How big is the device? My recollection is that the field
holding the device and partition size is a 32 bit signed
integer. With "only" 31 bits of usable value, that leaves
you with a per-device capacity maximum of 2,147,483,647
sectors (nearly 1 TB). The negative number you're seeing
probably means that you got into the space above 1 TB
but less than 2. Over two the integer will probably
quietly wrap.
On V4, the maximum supported UFS is 128 GB, though you
may be able to use larger (but not up to 1 TB in my
experience). AdvFS offers more options and with the
appropriate license you can make multiple volume domains
probably up to around 16 TB. Those limits are documented
either in the SPD/Quickspec for the operating system or
the release notes.
I sorted out the question of the missing LUN myself. What I thought were
two luns were actually two views of the first LUN (one through each of the
KGPSAs). That first LUN (/dev/rz8c) happened to contain a 1.3T volume,
hence the problem I reported. /dev/rzb8c (8764G) and /dev/rzc8c (534G) are
now labelled, in an AdvFS filesystem, and usable.
We're planning to upgrade to 5.1B in any case; this just speeds it up some.
Thanks again!
Ambar
Original query reproduced below:
> From: Jean Marie Diaz <jdiaz@westcoastbeauty.com>
> Date: 2003/09/23 Tue AM 11:44:48 EDT
>
> We are testing an EMC (Clariion CX400) connected via two KGPSAs to an
> AlphaServer 2100 running 4.0F patchlevel 8. The system sees two of the
> three LUNs (not sure why it only sees two):
>
> # file /dev/rrz17c
> /dev/rrz17c: character special (8/33794) SCSI #2 RAID 5 disk #136 (SCSI
> ID #1) (SCSI LUN #0) errors = 0/4
> # file /dev/rrz8c
> /dev/rrz8c: character special (8/16386) SCSI #1 RAID 5 disk #64 (SCSI
ID
> #0) (SCSI LUN #0) errors = 0/4
>
> I have edited /etc/ddr.dbase, compiled and rebooted as per the EMC
> documentation (also found in the archives here). I then zeroed the
> disklabel, and wrote a default disklabel, which when read back shows (note
> the negative number / overflow in the c partition definition):
>
> # disklabel -r rz17
> # /dev/rrz17a:
> type: SCSI
> disk: RAID 5
> label:
> flags: dynamic_geometry
> bytes/sector: 512
> sectors/track: 192
> tracks/cylinder: 1
> sectors/cylinder: 192
> cylinders: 8684
> sectors/unit: -1399230208
> rpm: 4400
> interleave: 1
> trackskew: 0
> cylinderskew: 0
> headswitch: 0 # milliseconds
> track-to-track seek: 0 # milliseconds
> drivedata: 0
>
> 8 partitions:
> # size offset fstype [fsize bsize cpg] # NOTE:
> values not exact
> a: 131072 0 unused 0 0 # (Cyl.
> 0 - 682*)
> b: 262144 131072 unused 0 0 # (Cyl.
> 682*- 2047)
> c: -1399230208 0 unused 0 0 # (Cyl.
> 0 - 15081963)
> d: 0 0 unused 0 0 # (Cyl.
> 0 - -1)
> e: 0 0 unused 0 0 # (Cyl.
> 0 - -1)
> f: 0 0 unused 0 0 # (Cyl.
> 0 - -1)
> g: 1447671936 393216 unused 0 0 # (Cyl.
> 2048 - 7542005)
> h: 1447671936 1448065152 unused 0 0 # (Cyl.
> 7542006 - 15081963)
>
> Creating an AdvFS file domain with the c partition then fails:
>
> # mkfdmn /dev/rz17c mirrortest
> mkfdmn: can't get device size; can't figure out file system partition
> mkfdmn: can't create new domain 'mirrortest'
>
> Is there some sysconfigtab variable or other bit of configuration that
needs
> to be set in order for disklabel to properly size the partition? We do
want
> to upgrade to 5.1B, but would rather get this working as is (our
production
> hosts are all at 4.0F).
>
> For extra credit, can anyone suggest a reason why we would only see 2 of
the
> 3 LUNs defined on the EMC?
>
> Thanks in advance,
>
> Ambar
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:36 EDT