Dieter Ries <[EMAIL PROTECTED]> posted [EMAIL PROTECTED], excerpted below, on Sat, 03 Jun 2006 14:05:30 +0200:
> and here is the output of emerge -e system: > > localhost / # emerge -ev system > Calculating system dependencies ...done! >>>> emerge (1 of 116) sys-devel/patch-2.5.9 to / [snip] >>>> sys-devel/patch-2.5.9 merged. > localhost / # emerge -ev system > Speicherzugriffsfehler > localhost / # So it emerges the first of 116 packages and returns to the prompt, no error or anything, then trying it again results in (courtesy babelfish) "memory access error"? Ouch! That single package merged, patch, yet replacing glibc fixes the problem? It makes no sense! If the problem is glibc, it shouldn't occur until you reach glibc in the emerge --emptytree cycle. Patch is simple and wouldn't be invoked by a second emerge attempt, that early in the attempt, so it can't be patch that's the problem. It has to be either portage (or python or bash, portage dependencies), or glibc, or your hardware (still). What happens if you "emerge -pe > pkg.list", to get a list of packages, and then emerge each one in order by itself (use --oneshot so you don't clog up your world file), going down the list one at a time? What about doing the quickpkg-off-the-CD thing with the packages listed above? Maybe your earlier issues caused one of them to emerge defective, and it's causing portage to do strange things. If so, once you replace the right one, you'll be back in business. I'd try python and portage first. as bash appears to be reasonably stable or your shell would be crashing and you'd be having to relogin all the time. It's also possible that your portage installed-package database is screwed up. Actually, this seems the most likely to me as it would explain a bit of the symptoms, being able to emerge a single package using --emptytree, which tries to write the new package info to the db, which then crashes emerge so it goes no further, and further attempts to emerge just crash. (There are still unanswered questions, but let's ignore them for the moment.) What's your disk layout? Do you have /var on a separate partition or is it on your main / partition? What filesystem are you using on it, ext3, reiserfs, something else? First, try copying /var/db/pkg to a different location for backup. Then delete it if possible. Unmount that partition if possible and do an fsck on it (maybe from a LiveCD if it's your root partition). Now boot or remount your system and NOW try an emerge -e system (-e = --emptytree). Does it work any better? You also mention errors scanning /etc, presumably using etc-update or similar. It may be that it needs fscked as well. If those don't work, it may be a lot of work, but it may be simpler to to think about starting from scratch again, with a new stage-3 install. Something's screwed up, and it's defying major attempts at finding and fixing it. I'm also beginning to wonder about the stability of your hardware. Someone recently posted with bad memory problems, and I've had both that and dying hard drives* in the last couple years, so I'm all too familiar with hardware issues. Good luck! I'm glad it's not me dealing with it, but you certainly have my sympathies! I've had close enough experiences in the past to know what it's like, and it's not fun! (BTW, quoting the context you are replying to, snipped from the entire message just to what you are replying to, then replying underneath, so a reader knows exactly what you are replying to, tends to work better for mailing lists and newsgroups than reply on top, unsnipped quote underneath. I know you have a lot to worry about right now, just sayin'.) * The hard drive thing was a case of a bad AC last summer causing drive overheating -- not surprising when summer temps here in Phoenix are known to routinely pass 46C (115F) and un-air-conditioned inside or sun baked temps can easily reach 60C (150F). The CPUs and rest of the hardware were fine, and the drive wasn't entirely dead either, but apparently had head-crash rings from the heat-expanded spinning platters contacting the head. I'm now running RAID-6 with Seagate drives with a 5 year warrantee, and have a new AC, so hopefully no more of that this year! -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman -- [email protected] mailing list
