Re: [gentoo-user] filesystem error?
On Tue, Jul 08, 2003 at 11:33:18AM -0700, Rex Young wrote: > Hello All, Hi, Rex. (Gosh what an unusual name :-) > I'm having a bit of trouble syncing. During an emerge sync I receive an > error: > > readlink dev-perl/Array-Window/Manifest: Permission denied > readlink dev-perl/Array-Window/files/digest-Array-Window-0.1: Permission > denied > readlink dev-perl/Array-Window/Changelog: Permission denied Can you post the output of: bash# id bash# df /usr/portage/dev-perl/Array-Window bash# ls -alh /usr/portage/dev-perl/Array-Window `type -p emerge` You say you received an error trying to chmod -- it would help to know the actual command you typed and the exact error you received. [Not including this kind of information is a pet peeve of mine, actually.] Regards, -- Rex -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] xargs:environment is too large for exec
On Tue, Jul 08, 2003 at 10:21:19AM -0700, Andrew Farmer wrote: > > The problem is probably that, when the command is run at boot, > (find /var/lock -type f -print0) returns a very long list of files, Hmm... Why? What would create a lot of files in /var/lock? How long is a "long list"? Hundreds of files? Thousands? Millions? xargs is reading from stdin!! It should be able to read an infinitely long list of filenames. > which is too long for xargs to deal with properly. xargs was written *explicitly* to get around the problem of "too many files passed". That's why it reads from stdin. The "-print0" option to find and "-0" option to xargs is a gnu extension (AFAIK) to allow for files containing newlines in their name, but xargs is definitely broken if it can't read an arbitrarily long list of null separated filenames from stdin. I'm still completely puzzled by this problem. I wish I could reproduce it without having to reboot! Regards, -- Rex -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] xargs:environment is too large for exec
> the line > > ebegin "Cleaning /var/lock, /var/run" > ( find /var/lock -type f -print0 | xargs -0 rm -f -- 1>&2 ) > > can be changed to: > > ebegin "Cleaning /var/lock, /var/run" > ( find /var/lock -type f -exec rm -f {} \; ) This definitely fixes the error but I'm curious why the first version doesn't work? Seems like a bug in gnu find or xargs to me -- the fix is really more of a workaround. The first version attempts to be more efficient (the latter forks a separate "rm" process for every file found). Can anybody explain what the problem is with the first version? When I try to run it manually (rather than as part of the runlevels boot process) it seems to work fine. Puzzled. -- Rex -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] mkinitrd...
On Thu, Jun 26, 2003 at 01:29:26PM -0400, [EMAIL PROTECTED] wrote: > 1) Is there some form of mkinitrd? I need to create a ramdisk to load some > modules (like reiserfs, evms, etc). However I've installed stage1 and stage2 > and I don't see the program anywhere. Hmm... I always just build the stuff I need for booting into my kernel rather than as modules, but I was surprised to see that mkinitrd doesn't seem to exist nonetheless. Fortunately mkinitrd is just a script. You should be able to just copy mkinitrd from another distribution (possibly requiring some edits). Alternately emerge lvm-user and use lvmcreate_initrd. Perhaps a better solution is the LFS mkinitrd referred to in this post: http://forums.gentoo.org/viewtopic.php?t=17695 > 2) Is there a way to search the portage tree for a paticular program? In this > case it would be nifty if I could search for mkinitrd even if it is in a > package that doesn't have the name in its title. I think the -s and -S options to emerge may be what you are looking for. > 3) I made a Raid 1 partition for /boot. Can i do this and boot with grub? I don't think so (md needs the linux kernel), but why even try? Gentoo wisely leaves /boot unmounted. I have two drives in my system, /dev/hda1 is my /boot partition. I created /dev/hdb1 the same size as /dev/hda1 and just use "dd if=/dev/hda1 of=/dev/hdb1 bs=8192b" to back it up (while unmounted). Kind of a poor man's raid1, but more than sufficient for backing up something that changes as rarely as /boot (only takes a few seconds to back it up, too). [NOTE: wisest to only do your backup immediately AFTER succesfully booting a new kernel!] I'm in the process of putting together a disk-based backup scheme (tapes are too much hassle, and drives are pretty cheap these days). I use LVM for everything except /boot so I can shrink/grow on demand. (Actually, I've still got root on /dev/hda5, but I plan to move it to a raid1 volume REAL SOON NOW... :-). I've got drive space to burn (two 80GB drives) and wanted the performance of raid0, but I also wanted to be able to survive a drive failure: so I backup my raid0 volumes to (normally unmounted) raid1 volumes. 3X the drive space normally required (at least for the stuff I backup) but, hey, it's darn fast. :-) Once I get things set up the way I like with completely automated backups (and successfully simulate a drive failure and recovery) I'll post my experiences to the documentation forum. Regards, -- Rex -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] DMA and ide-scsi
On Thu, Jun 26, 2003 at 05:33:35AM +0100, Richard Revis wrote: > > I have found references in several places which state that you should be > able to use hdparm to enable dma on ide-scsi devices, however when I > enable ide-scsi the /dev/hdX entry (in this case /dev/hdd) for it vanishes > and hdparm refuses to work when pointed at the SCSI entry. Any clues as to > what is happening? Google found this for me: http://www.geocrawler.com/archives/3/35/2001/2/0/5251986/ In a nutshell you may need to run hdparm before loading ide-scsi. Regards, -- Rex -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] Problem with scp
On Wed, Jun 25, 2003 at 07:48:49PM +1000, Paul Colquhoun wrote: > > On Wed, 25 Jun 2003, Paul Colquhoun wrote: > > > > >The only problem I am having is with scp. I have openssh installed, > and > > >can ssh into the box, but cannot copy files to the box with scp. When > I > > >try, I get asked for my password (as expected), then the session just > > >sits there, doing nothing. [snip] > $ scp xx [EMAIL PROTECTED]:. > [EMAIL PROTECTED]'s password: Please post the output of "scp -v xx [EMAIL PROTECTED]:" It's probably your shell trying to be interactive (are you using csh/tcsh by any chance?). Try temporarily renaming all shell configuration files (~/.{cshrc,login,profile,bash_profile,bashrc). Regards, -- Rex -- [EMAIL PROTECTED] mailing list