Note: This bug is displayed in read-only format because the product is no longer active in Red Hat Bugzilla.

Bug 6463

Summary: kudzu doesn't rescan scsi bus after detecting controller
Product: [Retired] Red Hat Linux Reporter: Bob Drzyzgula <bob>
Component: kudzuAssignee: Bill Nottingham <notting>
Status: CLOSED WONTFIX QA Contact:
Severity: medium Docs Contact:
Priority: low    
Version: 6.1CC: bob, rvokal
Target Milestone: ---Keywords: FutureFeature
Target Release: ---   
Hardware: i386   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Enhancement
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2005-02-04 22:40:56 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Bob Drzyzgula 1999-10-28 14:41:51 UTC
Hi. I installed RH6.1 from a CD burned from the ISO image
downloaded from your site a week or so ago. The original
build only had an AIC7895 SCSI Controller (Tyan Thuder
Dual, with 440LX chipset). I later installed a DPT PM3334UW
(I was having trouble with an external chain off the
Adaptec, which I must say has been no end of trouble to
me...).

What happened was that while kudzu properly identified the
new device as a DPT controller, its pci mapping table
specifies the wrong driver; it lists "dpt", when there is
no module called "dpt". The proper entry should be "eata".
Once I changed that, kudzu worked properly.

The other problem I had was that, since your setup by
default loads the SCSI driver modules in the initrd, and
since when a new controller is added the initrd isn't
updated, kudzu had trouble with the drives attached to the
controller. If I ran kudzu by hand and then rebooted, kudzu
would be happy about the DPT controller (as long as I had
fixed the pcitable), but it would tell me that the drives
had been removed from the system. Once I edited the initrd
to autoload the eata driver as well as the AIC driver,
everything was fine and dandy.

So, I think that the things that need to be fixed are (a)
the pcitable, and (b), ultimately I think that if kudzu
identifies a new SCSI controller you should be providing a
facility to automatically update the initrd (assuming that
you are trying to avoid the need for people to compile new
kernels.).

Following is a link to a thread on Deja.com, where I
reported this problem in case others experience the same
thing; there's a little more detail in there:

http://www.deja.com/viewthread.xp?AN=541265448

Thanks,
--Bob Drzyzgula
bob

------- Additional Comments From   10/28/99 10:43 -------
(Note that I also sent this to bugs, and came to
file it here because of the "deprecated" autoresponse from
the former address...)

--Bob

Comment 1 Bill Nottingham 1999-10-28 15:25:59 UTC
The pci id mapping problem is fixed in the latest kudzu in
Raw Hide.

The second one is something we definitely need to work on
clearing up...

Comment 2 Bob Drzyzgula 1999-10-28 19:10:59 UTC
(Note that I also sent this to bugs, and came to
file it here because of the "deprecated" autoresponse from
the former address...)

--Bob

Comment 3 Brent Fox 2002-06-05 01:12:01 UTC
notting, do we update the initrd correctly now?

Comment 4 Bill Nottingham 2002-06-05 01:31:55 UTC
No, mkinitrd is not re-run...

Comment 5 Bill Nottingham 2005-02-04 22:40:56 UTC
Closing out bugs from older, no longer supported releases. Apologies for the
lack of response.