[Rule-list] Miniconda dump available online - UPDATED: Succesful Installation

[C David Rigby]

On Thursday 26 September 2002 08:31 am, C David Rigby wrote:
> On Thursday 26 September 2002 12:44 am, Michael Fratoni wrote:
> > -----BEGIN PGP SIGNED MESSAGE-----
> > Hash: SHA1
> >
> > On Wednesday 25 September 2002 12:39 pm, C David Rigby wrote:
> > > Cool.  I noticed also that there are some things that I didn't do per
> > > the instructions (yes, I have heard of such things as manuals, but...
> > > {8->) so I will go back and try this a bit differently before I start
> > > adjusting hardware, etc.
> >
> > Hi David,
> > When you try again (with the instructions in hand) ;)
> > Let me know how it goes, please. There have been a lot of reports on the
> > Valhalla list for Red Hat 7.3 reporting segfaults during install on older
> > hardware, and I haven't had any time to look at this with regards to the
> > miniconda installer. The miniconda installer should behave much like the
> > normal Red Hat install, only in text mode, with fewer options. It does
> > use the same kernel, modules, rpms, and build process as the standard
> > anaconda installer.
> >
> > Which Red Hat version are you trying to install? Miniconda v-0.7.35 and
> > v-0.7.4 are for the 7.3 release, Versions <= 0.7.3 are for the 7.2
> > release. That right there could account for some of your difficulties.
>
> mincondas 0.7.4 and 0.7.3x fail too early in the process (during
> /sbin/loader) for the CDs to be accessed.  I only really get into the
> installer with miniconda-0.7.2 and then I use the RH7.2 CDs.
>
> > Since I've been away from this for a while, I'm going to attempt to find
> > time this weekend to put together another build of miniconda.
> >
> > - --
> > - -Michael
>
> Michael,
> 	I have tried several more times without success, in the following
> fashions. You may want to just skip to the bottom to see my conclusion
> before plowing through all these details.
>
> 1)
> 	Discovering that I needed the updates diskette (this was the reading
> instructions part) I attempted the miniconda-0.7.2 install again.  I did
> reach the point where I was presented with the installation options,
> including the Low Memory system, which I had not seen before.  After
> resolving dependencies, the installer crashed again in a very similar
> fashion (I have another anacdump.txt available from this install attempt,
> if someone would like for me to send it).
>
> 2)
> 	I tried again, this time the difference being that I chose a check for bad
> blocks during partition formatting.  Recall that I had previously used the
> RH6.2 installer to access fdisk and partition the drives, and they were
> recognized by Disk Druid under miniconda-0.7.2 as type "linux native" and
> not as ext2.  I made it to the same point as (1) above, but this time the
> failure was a bit different.  On the main console I got the message
> 	Message: An error occured while searching for bad blocks on sda1 (this is
> my /var partition).  This problem is serious, and install cannot continue.
> Other consoles showed the following:
> 	F4 showed lots of "Error -3 while decompressing"
> 	F5 showed "badblocks: error while loading shared libraries: badblocks:
> symbol free, version GLIBC_2.0 not defined in file libc.so.6 with link time
> reference"
>
> 3)
> 	As a sanity check, I tried RH6.2.  It also failed in the same fashion as
> the anaconda failures that provided the previously mentioned dumps, at
> about the same point, which was after chosing the type of installation.  It
> did not provide an option for a dump, only to debug.
>
> So now I am suspecting some sort of subtle hardware problem, since
> installing RH6.2 in 32MB of RAM has never been an issue before.  The
> machine is almost ten years old, afterall.
>
> 4)
> 	Trying to simplify things, I removed the external 2GB Fujitsu SCSI HDD,
> and tried miniconda-0.7.4 again.  But it still failed as previously:  the
> last console message from the installer is after the ramdisk loads, and
> after the line:
> running /sbin/loader
> one gets the message:
> "install exited abnormally -- received signal 11"
> and then the system shutsdown gracefully.
>
> 5)
> 	With the external SCSI drive removed, I tried miniconda-0.7.2 again, with
> bad block checking enable for the partition formatting, and the installer
> failed at the "reading packages" stage with a signal 11 abnormal
> termination.  Hmmm. That's a new one.
>
> 6)
> 	Still without the external SCSI drive, I tried miniconda-0.7.2 again,
> planning to forego the bad block checking.  The process abnorally
> terminated on signal 11 again, well before reaching the Disk Druid portion
> of installation.
>
>
> CONCLUSION
> ----------------
>
> Before we spend a lot of time debugging the software, it's time to verify
> the hardware.  The failures are not following a reasonable, reproduceable
> pattern.  So, I will:
> 1) Remove the SCSI controller and NIC, and disconnect the SCSI CD-ROM
> drive.
>
> 2) Run some testing software I have against the hardware for a 24-hour
> period. It will give the motherboard, memory and IDE hard drive a thourough
> workout.
>
> 3) Re-initialize my diskettes and recreate them.
>
> 4) If I am still unable to install at that point, there are a couple of
> other tricks available: I have yet another Adaptec 1522A controller, and I
> have also some other SCSI controllers to try.  If all else fails, I can try
> again with an IDE CD-ROM drive in place of the SCSI drive.  Of course,
> there are other old PCs as well...yes this place is a real computer junk
> yard!  {8->
>
> It will take a while to carry out all of this, and I have an exam coming up
> so I need to pay some attention to school.  But I will summarize the
> results once I have had a chance to run through these tasks.
>
> CDRigby
> cr8s at virginia.edu


I have succeeded in installing RedHat 7.2 in the RULE Low-Memory configuration 
via miniconda-0.7.2.  Here is what was needed:

1) hardware changes
	The most necessary change for this old machine seems to have been replacing 
the 1046 cylinder drive with a smaller drive that did not exceed the old 
BIOS's limit of 1024 cylinders.  Previously, standard RedHat 6.2 installation 
would not complete.  After the hard drive swap, it did.
	It should also be noted that the external 2GB SCSI hard drive was detached 
for all that follows.
	I also replaced the floppy drive (a weak eject spring - I got tired of 
removing floppy disks with a pair of needle-nose pliers) and the power supply 
(sometimes the system would not start up).
	Still, miniconda-0.7.2 would fail as before: after selection of the 
Installation Type and before the actual installation of packages had begun.
	Finally, I swapped out the AHA-1522A SCSI controller and tried again with a 
Future Domains TMC-1660ER (installer could not insert module), a Trantor T128 
(timeouts killed installation) and finally an Adaptec-154x-based card.  With 
this last card in place, installation completed succesfully.

2) Debugging
	If it is useful, I have a dump that I can send to whomever would like to see 
it that was the output after failure of the last attempt at installation 
using the AHA-1522A.  It may be helpful in debugging support for this 
relatively common old SCSI card.

3) To do
	First, retry the installation again with the 2GB external SCSI drive 
attached.  Current available storage on the internal HD is limitted to 320MB 
for /, and the installation of the base Low-Memory system w/ Network and DHCP 
Client support used 120MB, and the installer indicated it needed 302MB for 
the installation.  So, more storage would be nice.
	Second, retry the installation on another PC (I have an older 1st generation 
P5 around here somewhere) again using the AHA-1522A.  This should give an 
indication of whether or not the problem is really this particular card, or 
some sort of interaction between the ancient 486 and that particular SCSI 
card.
	Also, try installation on a different computer with the AHA-1522A using 
miniconda-0.7.4, which still fails early in the process.  The kernel loads, 
the ramdisk loads, and then failure occurs while running /sbin/loader with a 
signal 11.


_______________________________________________
Rule Project HOME PAGE:  http://www.rule-project.org/rule/
Original Rule Development Site http://savannah.gnu.org/projects/rule/Rule-list at mail.freesoftware.fsf.org
http://mail.freesoftware.fsf.org/mailman/listinfo/rule-list