Red Hat Bugzilla – Bug 33513
Install via NFS on 486 fails
Last modified: 2007-04-18 12:32:25 EDT
Trying to install on my 486 with bootnet.img using NFS, I get passed DHCP and NFS names and then the last thing I see is: "Running anaconda - please wait" then I get: install exited abnormally -- received signal 11 sending termination signals...done disabling swap... unmounting filesystems... /mnt/runtime /mnt/source umount failed (16) /dev/pts /proc you may safely reboot your system The previous version of rawhide "nearly" installed on this machine (packages got installed, but the installer bugged out at the very end, which is also why I'm trying this slightly newer version). Maybe an anaconda issue?
Does it have a PS/2 port? How much memory does it have?
Machine config: Passive backplane mobo (ISA only) AMI BIOS 080893 486 66MegaHurt, have also tried with a Pentium Overdrive to no avail... 32MB mem 1GB IDE HD 3Com EtherLink3 (3c509) CirrusLogic GD5424 1MB? PS/2 keyboard Serial mouse
What's the first message *before* the segfaut?
Hmmm, you mean between "Running anaconda" and "exited abbynormal"? didn't see anything... does a clear screen just when the sig11 hits. Got any debug flags for the installer so I can see what's really going on?
Are you trying to run the installer in GUI or text mode?
installer run in "text" mode...
Did you try Wolverine at any point? As you know, Rawhide is a test bed and as such, it may be less than stable. Sounds like this could be a kernel or a hardware issue, but that's just a guess. Sorry, there are no super secret debug flags for anaconda. Most of the debug messages are logged on VC3 and VC4. So try this...as soon as you hit OK on the screen where you specify the NFS server, hit Alt-F3 and see if any messages appear out of the ordinary. Try Alt-F4 too. The problem here is that once you get the sig 11, you may not be able to go back and forth between VC's anymore, so you may have to try this more than once to see the messages on both VC3 and VC4. Do you see any error messages?
Checked VC3 and VC4, last thing that shows up: VC3: * looking for USB mouse VC4: <6>raid5 personality registered as nr 4 I presume that the USB mouse load comes after the raid... and there is no USB in this PC... History: First I tried RH7.0 which didn't work, so I tried Debian 2.2 to check if the PC would run Linux at all and it installed flawlessly, but being a RedHat afficianado I wasn't giving up so easily. So I moved on to Wolverine, which has a known kernel bug that broke 486 boot. So then I tried RawHide (the previous version) which nearly made it to the end of install (really close, past package installation etc) but the next day there was a new release of RawHide so I downloaded that and here I am...
Now I am confused. We know that Red Hat 7 works with 486s...and we've tested our current internal builds too. Normally I would think this is a hardware problem, but since Debian works, I guess that's not it. So do the newer trees fail in the same place? Has there been any improvement at all?
Newer rawhide trees would not make any difference. The installer in rawhide hasn't been updated in a few weeks.
Downloaded new RawHide thursday last, installation now passes without errors. Thanx anyway guys...
Oops, yeah, I was on crack. New rawhide wouldn't make a difference because the installer code is the same, but it *would* make a difference because the boot kernel the installer is running is different. So it's apparently a fixed kernel bug. :)
Marking as 'fixed in 7.1'.
(hm, or I suppose it could have been the kudzu mouse probe bug, but that shouldn't have segfaulted, even with the previously buggy code, as long as there was a ps/2 port on the machine.)