home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!wupost!sdd.hp.com!mips!odin!fido!zola!twilight!zuni!anchor!olson
- From: olson@anchor.esd.sgi.com (Dave Olson)
- Subject: Re: Formatting 2 gigabyte disks (also, striping questions)
- Message-ID: <nkp4kdg@zuni.esd.sgi.com>
- Sender: news@zuni.esd.sgi.com (Net News)
- Organization: Silicon Graphics, Inc. Mountain View, CA
- References: <1992Jul22.213912.13138@ncsu.edu>
- Date: Thu, 23 Jul 92 00:51:51 GMT
- Lines: 71
-
- In <1992Jul22.213912.13138@ncsu.edu> garvin@cfd.mae.ncsu.edu (Michael Garvin) writes:
-
-
- | I have what are probably a few senseless questions. Our setup is a
- | Power Series 4D/440 with a Jaguar differential SCSI controller. Connected to
- | each channel is a third-party drive tray containing four Micropolis 1548 disks
- | (2,000 MB/unit unformatted capacity, 1,748 MB/unit formatted) and a single
- | ended SCSI to differential SCSI converter. After formatting and partitioning
- | the disks I notice a few curious things. First, fx reports, correctly I
- | believe, the drive geometry:
- ...
- | showing 2,112 cylinders which is close to what the Micropolis manual states
- | (2,099 unformatted, 2089 formatted). However, the partition table (partition
- | ten in particular) tell a different story:
- |
- | ----- partitions-----
- | part type cyls blocks Megabytes (base+size)
- | 0: efs 3 + 16 5871 + 31312 3 + 15
- | 1: rawdata 19 + 41 37183 + 80237 18 + 39
- | 6: efs 60 + 1651 117420 + 3231007 57 + 1578
- | 7: lvol 3 + 1708 5871 + 3342556 3 + 1632
- | 8: volhdr 0 + 3 0 + 5871 0 + 3
- | 10: entire 0 + 1711 0 + 3348427 0 + 1635
- |
- | Is this discrepancy (2,089 vs 1711 cylinders) a possible bug in fx? What leads
-
- No. The 'cylinder' numbers in the partition layout are logical cylinders,
- calculated to give the maximum capacity on the drive. You will something
- similar to this with most zoned drives, because sectors/cylinder isn't
- a constant.
-
- | me to think this is that 1,711 works out close to 1.65 gigabytes, the size of
- | the largest drive SGI is supporting. I've tried altering that parameter, but
- | fx refuses to let me do so.
-
- You can use the /debug/showcapacity menu to show the actual capacity
- (in sectors) of the drive. It should be quite close to 3348427, but
- somewhat greater. Perhaps the formatted capacity you were quoted doesn't
- include all the spare sectors (8*2089 + 45*131 is another 22607 sectors,
- but that still doesn't bring you up to 1.748 Mbytes). It is also possible
- that they are quoting 10^6 Mbytes, instead of 2^20 Mbytes; I tend to
- suspect this because 1.748 Gbytes (10^9 Gbytes) gives 3334000 sectors,
- which is pretty close to what you show.
-
- | Secondly, I was wondering if someone would critique my striping scheme.
- | Here's my /etc/lvtab file:
- |
- | -=-=-=-=-
- | #
- | # Logical volumes on Falcon drive trays (each with four two-gigabyte drives).
- | #
- | lv0:First 8 Gigabyte Striped Logical Volume:stripes=4:devs=/dev/dsk/jag0d2s7, \/dev/dsk/jag0d3s7, /dev/dsk/jag0d10s7, /dev/dsk/jag0d11s7
- |
- | lv1:Second 8 Gigabyte Striped Logical Volume:stripes=4:devs=/dev/dsk/jag0d4s7, \
- | /dev/dsk/jag0d5s7, /dev/dsk/jag0d12s7, /dev/dsk/jag0d13s7
- | -=-=-=-=-
- |
- | As I understand it this should provide maximum performance by striping across
- | each of the four two-gigabyte drive (an eight gigabyte partition), with two
- | drives in each logical volume being on a controller channel. Comments?
-
- I haven't worked with the jag controllers much, but I don't think this
- is really the ideal configuration, since 2 of the drives in each lv are
- likely to be idle at any given time (for sequential i/o).
-
- Specifying a step size may help.
-
- --
- Let no one tell me that silence gives consent, | Dave Olson
- because whoever is silent dissents. | Silicon Graphics, Inc.
- Maria Isabel Barreno | olson@sgi.com
-