From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 5.01; Windows NT 5.0)
Description of problem:
Gets 98% of way thru package installation, fails with signal 9
(anaconda.txt dump produced). Environment: Text mode install, Buslogic BT-
946C scsi controller, installing to second disk. No IDE drives; scsi card
has 2 HDs and a plextor CDROM. CPU P166, RAM 59 MB. Video ATI Mach64 (have
to use text mode). Installing on 2nd disk (ID=1), 2.1 GB, dedicated
(nothing else on the drive). Used DD to partition disk: 16 MB boot, 256 MB
swap, rest is root. Tried various install modes: custom, server, etc.,
with minimal package selections. Same result. No info in VC's.
Steps to Reproduce:
1. Boot from diskette (boot.img)
2. Partition SCSI drive (sdb1, sdb5, sdb6) using Disk Druid
3. custom install, select hardware & packages
4. begin install
Actual Results: Progress bar almost complete, get internal error. Crashes
with signal 9.
Expected Results: Installation completes, writes image, etc.
May be related to Bug # 39604. Note: both entail machines with
less than 64 MB, and the comments there point to a patch for this
Created attachment 21405 [details]
Did you set mount points for any DOS partitions during the install?
Also, did you go on VC2 and enter the /mnt/source directory?
6/21/01 No, I didn't set any Dos mount points. But I later ran a totally
minimal install -- no packages -- and it was successful. At one point the
package installer choked while loading some clib libraries. After three tries,
it succeeded in installing the file. I checked the VC3 log, and it identified
an error on SCSI 0 (the only SCSI controller), ID #6, which is the CDROM unit.
It appears from all the information that the install program is encountering
occasional IO errors on the CDROM unit -- an old Plextor 6X. I suspect this
unit may be flaky and may be causing intermittent problems.
Also noticed a message in one of the VCs that it "saw" 64MB. Interesting,
because there are only 60 MB RAM on the machine. Didn't seems to cause a
Anyway, I'll close this bug on the belief that IO errors on the CDROM unit
are the cause.
Do you see anything on VC4 that has anything to do with DMA errors, or CDROM
6/22/01: I don't recall seeing anything about these on VC4, but I'll look for
them when I run the rebuild.