Mailing List Archive


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [tlug] Master and slave



On Monday 19 May 2003 13:48, Michael Engel wrote:

First comment: I have never experienced such a problem [1], and
if I have a system with one disk and one CD, I always put them
as masters on different IDE headers.

> connected to the EIDE controller in the correct order.

EIDE = ATA-2 and != ATA-Ultra/anything, so I wonder if
SuSE is referring in particular to a problem involving EIDE
and Linux?

In addition to the handy URL posted by J-C, the links below
are also useful:

http://en.tldp.org/HOWTO/SCSI-2.4-HOWTO/scsibus.html
http://www.interfacebus.com/Design_Connector_IDE.html

Where the SuSE manual says:

> (secondary IDE controller). This is wrong and can cause Linux not to know
> what to do with this "gap".>

it may be that they are dicussing a bug that was specific to (some version
of) the Linux kernel or a relevant driver.  If so, then it is not the hardware
config that is wrong.  In any case, as I noted above, I have never seen
this problem.  If you haven't either, don't worry about it and ignore what the
SuSE manual says.

I also question the manual b/c of what I wrote previously about RAID.  If
you are running ATA RAID (whether hardware or software) and you 
put two disks on the same header, you will kill your performance.  The
best setup is first RAID disk on first ATA header, second RAID disk on
second ATA header [2], and CD somewhere else [3] [4]

BTW, if anybody has a link proving the manual right and the rest of us
wrong, by all means please post it.

Footnotes
========
[1] On Linux, FreeBSD, or Windows.
[2] RAID disk here refers to a physical disk, as opposed to
    a logical one (RAID volume).
[3] Either SCSI, as I have it at home right now, or
    a third ATA header, as I have it at work, or USB/Firewire
    if  you have such a CD drive.
[4] Note that if you have a setup with a disk on header one,
    a disk on header 2, the drive letter assignments will be
    /dev/hda and /dev/hdc, because IDE does number the devices
    such that the first device on header 2 is device 3 (device 2
    is the second device on header 1).  However, there doesn't
    *have* to be a device 2 as far as my experience and googling can
    determine, so I would call it a bug if something has a problem with
    a setup containing devices 1, 3, and 4 but not 2.  Again, if anyone
    can prove my wrong, please do so.

Jonathan Q
GPG key ID: ACC46EF9 (E52E 8153 8F37 74AF C04D  0714 364F 540E ACC4 6EF9)
To get my public key: gpg --recv-keys --keyserver pgp.mit.edu ACC46EF9

Attachment: pgp00055.pgp
Description: signature


Home | Main Index | Thread Index

Home Page Mailing List Linux and Japan TLUG Members Links