Re: Redhat-install-list digest, Vol 1 #1815 - 13 msgs

["Steve Cousins" <cousins umit maine edu>]

Jon Adam writes:
>
>This is not entirely accurate on your end either.  It depends on the brand
>of CDR you are using.  I'm sure Rick Stevens will agree with me
>here.  Some CDRs, the outer edge is unreadable, thus leaving off some of
>the necesary RPMs and such.  In my opinion, if it came from Mr. Stevens,
>it is never false.  He has always steered us right and helps many people
>on the list.  I would pretty much bet you if you would have used 700 meg
>CDRs, the first try would have went right through.

>>
>> The suggestion that you need 700 MB CD's is false.  I was successful
>with
>> 650 MB CD's. There is no reason at all that you need 700 MB CD's.  The
>> images are at most 638 MB.
>>

My statement was entirely accurate.  You don't need 700 MB CD's.  I have a set of working Redhat 7.3 650 MB CD's.  That you need good quality CD's is not disputed.  I understand that maybe you get a better chance of success with 700 MB CD's but the
blanket statement that if you use 700 MB cd's you will not have any problems is false.  I made a set of 700 MB CD's and had the same problem with disk three.

I stress this point because it takes away a variable.  By doing this, it helps to see what the real problem is.  

The response to this problem has been:

	1. Use 700 MB CD's
	2. run md5sum on the ISO's before burning
	3. run media check on the CD's after the burn.

As stated above, while you can use 700 MB CD's, it doesn't guarantee a good set of CD's. 
Step two is good advice.  You don't want to burn a bad image.
Step three seems to be useless. I ran it on all of my CD's and it always said that the CD's were fine, even though they weren't always fine.

The real problem seems to be a combination of things:
	
	1. something odd with the CD burn
	2. mediacheck gives false positives

Does anyone from Redhat read these posts (it seems not)?  If so, I would recommend that they add something to mediacheck to go through the same check that the install process does to see if the CD's are correct.  That is:

	1. mount the CD
	2. check for the .disc?-<arch> file
	3. read the timestamp from the first line of the .disc* file
	4. compare it with the timestamp of the CD that mediacheck is running from

If there are errors on any of the steps, report them to the user with a useful suggestion as to what the problem is.  All of them except for step four point to a bad burn.  Step four points to an inconsistent set of ISO images.  

The answer for me was to keep on burning CD's until it worked.  I used CDRW disks so it wasn't nearly as wasteful.  I was installing to a seperate machine from the one that my CD burner was in so when the installer complained about the CD not being
the correct one, I would take the CD out reburn it, and then put it back in.  The second burn for each CD worked.  I was also very careful to not do anything CPU intensive on the machine while it was reburning the CD.  Possibly this was the
difference.

bkortiak macsteelusa com writes:
>
>I'm pretty sure he did try 700MB CDR's without success (or maybe that was
>another thread).  While many of the problems could be eliminated with the
>700MB variety, the specific problem he seems to be describing, and its
>resolution, would not have been alleviated by the simple expedient of using
>media of greater capacity.

Exactly!
________________________________________________
 Steve Cousins                        Email: cousins umit maine edu
 Research Associate                 Phone: (207) 581-4302
 Ocean Modeling Group           
 School of Marine Sciences      208 Libby Hall,
 University of Maine                Orono ME 04469