Re: [gentoo-user] syslog-ng segfaults
On Fri, 19 Jul 2013 17:03:36 -0400 Randy Barlow ra...@electronsweatshop.com wrote: Alexey Mishustin wrote: So, restarting syslog-ng should be all that's required to fix it - reboot is overkill. As for me, first I updated syslog-ng, then I issued '/etc/init.d/syslog-ng reload' (by mistake, instead of 'restart'), and then 'restart' as I should. Then, just when syslog-ng was restarting, the segfault happened. I also noted that restarting syslog didn't seem to solve the problem. I do think Adam's reasoning makes sense, but there must be something else that needed to be restarted as well. Same behaviour here. In my case with an lsof | grep libsyslog-ng I see in the physical host hp-systray from hplip was still using the old libsyslog-ng.so, so killing that and a restart of syslog-ng service stops the segfault lines. YMMV, Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Re: Udev update and persistent net rules changes
On Sat, 6 Apr 2013 19:11:46 +0700 Pandu Poluan pa...@poluan.info wrote: On Apr 6, 2013 3:44 PM, Neil Bothwick n...@digimed.co.uk wrote: On Fri, 5 Apr 2013 21:14:39 -0400, Walter Dnes wrote: * on a machine with multiple network cards *ALL USING DIFFERENT DRIVERS* * drivers are built as modules, not built-in into the kernel * is it possible to set things up so that the network driver modules do not load automatically at bootup? * have a script in /etc/local.d/ (or wherever) modprobe the drivers in the desired order I can see complications involving services that depend on net (e.g. sshd), but in general, would it work reliably? What happens if one of the modules fails to load for any reason? If you need persistent device names, set up rules to give them, but use names outside of the kernel namespace to avoid kk problems that udev is trying to avoid with its new naming rules.ooh Ahhh... I think now I understand... So. Here's my summarization of the situation: * The ethX naming can change, i.e., the interfaces can get out of order * So, to fix this, udev decided to use the physical attachment points of the NIC in driving a persistent name, a name that will be identical across boots as long as there is no hardware change There are also other ways such as using the mac address (disabled by default). * In doing so, it also frees the 'traditional' ethX names to be used No. The eth[0-9]+ namespace is not free, it has always been used by the linux kernel, and will stay so. * If one wants, one can still 'rename' the NICs to the 'traditional' names using the 70-*.rules script * Doing so (specifying the NICs' names using the 70-*r.rules script) will also disable the new 'persistent naming' system -- for the NICs specified in the 70-*r.rules file * Therefore, users that will be impacted are those that upgraded udev but doesn't have the 70-*r.rules, for udev will then assign new names for the NICs * For these users, specifying the netsomething switch for the kernel (sorry, forgot the complete switch) will disable the new naming system So, have I gotten everything correctly? Almost, except you should not specify a name that is also eth[0-9]+ (what you called 'traditional' name), since it can cause a race condition where the kernel and udev fight for the name. While it used to be the case (i.e. udev-197) that udev tries to handle the race condition, the devs has decided to remove those code. Regards, Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] /sbin/ldconfig: /usr/lib/libgmp.so.3 is not a symbolic link
On Wed, 13 Feb 2013 03:43:52 + Mateusz Kowalczyk fuuze...@fuuzetsu.co.uk wrote: For a longer while now I've been getting `/sbin/ldconfig: /usr/lib/libgmp.so.3 is not a symbolic link' warning every time I build something. It didn't seem to create any actual issues and a quick Google search didn't bring anything up so I've ended up ignoring it. I am now sick of it and would love it fixed. I can't find anything on the web with anyone having the same problem and rebuilding libgmp seems to have had no effect. Does anyone have some ideas as to how the issue can be fixed? Thanks, Mateusz Kowalczyk (Delete the file /usr/lib/libgmp.so.3 and) check for invalid symlinks? Surely /usr/lib/libgmp.so does not point to /usr/lib/libgmp.so.3.whatever, but some more recent .so version, like /usr/lib/libgmp.so.10.0.2? Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Re: Heads up if you start X with startx; xorg-server suid flag
On Mon, 31 Dec 2012 10:03:40 +0200 Alan McKinnon alan.mckin...@gmail.com wrote: It's not in the profile, the xorg-server ebuild sets USE=suid on by default. Most likely is that Walter has USE=-suid in his make.conf and sets it back on for things he's checked out personally. Meaning that in this case one slipped through. I suspect it is a USE=-* (blah) rather than an explicit USE=-suid in the make.conf file. One question though --- should the xorg-server ebuild be such that IUSE=(blah) +suid when using a hardened-profile? Also, checking my PORTDIR, given the global description in use.desc (suid - Enable setuid root program, with potential security risks), shouldn't the suid use flag entries (net-analyzer/nagios-plugins:suid and net-wireless/kismet:suid) be deleted from use.local.desc? Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Re: Heads up if you start X with startx; xorg-server suid flag
On Mon, 31 Dec 2012 11:29:12 +0200 Alan McKinnon alan.mckin...@gmail.com wrote: On Mon, 31 Dec 2012 16:53:47 +0800 kwk...@hkbn.net wrote: On Mon, 31 Dec 2012 10:03:40 +0200 Alan McKinnon alan.mckin...@gmail.com wrote: It's not in the profile, the xorg-server ebuild sets USE=suid on by default. Most likely is that Walter has USE=-suid in his make.conf and sets it back on for things he's checked out personally. Meaning that in this case one slipped through. I suspect it is a USE=-* (blah) rather than an explicit USE=-suid in the make.conf file. One question though --- should the xorg-server ebuild be such that IUSE=(blah) +suid when using a hardened-profile? That already has a de-facto answer; USE=suid must be on by default as without it users cannot run a desktop (xorg-server does not yet run without root permissions) But(!) if one uses a login manager, xorg server would only be ever be run by root, right? Hence the use flag rather than a must like, e.g., sys-apps/shadow (and the question whether the dangerous suid should be set in desktop profiles instead of default on even for hardened). Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] emerge --update : how to keep it going?
On Sun, 2 Dec 2012 16:12:02 +0200
Alan McKinnon alan.mckin...@gmail.com wrote:
On Sat, 01 Dec 2012 19:58:45 +
Graham Murray gra...@gmurray.org.uk wrote:
Volker Armin Hemmann volkerar...@googlemail.com writes:
--keep-going does not help you, if the emerge does not start
because of missing dep/slot conflict/blocking/masking whatever...
Though it would be nice if there was some flag, probably mainly of
use with either ' -u @world' or --resume, to tell portage to get on
and merge what it can and leave any masked packages or those which
would generate blockers or conflicts.
That is a terribly bad idea, and you need to have a fairly deep
understanding of IT theory to see it (which is why so few people see
it). I don't know which camp you are in.
The command is to emerge world, and it's supposed to be determinate,
i.e. when it's ready to start you can tell what it's going to do, and
that should be what you told it to do, no more and no less[1]
the command is
emerge world
not
emerge the-bits-of-world-you-think-you-can-deal-with
If portage cannot emerge world and fully obey what root told it to do,
then portage correctly refuses to continue. It could not possibly be
any other way, as eg all automated build tools (puppet, chef and
friends, even flameeyes's sandbox) break horribly if you do it any
other way. Life is hard enough dealing with build failures without
adding portage do somethign different to what it was told into the
mix.
[1] determinate excludes build failures, as those are not
predictable. Dep graph failures happen before the meaty work begins.
While there are good reasons not to implement it in portage itself, you
can implement it with a bit of help from shell scripts telling portage
what to do.
Do an emerge -uDpv world, use sed or awk or whatever to replace the
beginning [ebuild ...] and whatever come after the package
nameversion, and finally loop emerge -1 =${package} for each package
in that list. Now provided you discard the return value of emerge, if
such ${package} will give you something that portage doesn't think is a
good idea (e.g. unsatisfiable dependencies), the loop will go on to the
next package instead of completely halting.
The shell script is thus left as an exercise.
The usual warning applies:- it can be done doesn't necessarily mean it
is a good idea to do it.
Kerwin.
signature.asc
Description: PGP signature
Re: [gentoo-user] [Offtopic] Lightweight server distro for an old motherboard
On Wed, 29 Aug 2012 10:57:07 +0800 Andrew Lowe a...@wht.com.au wrote: Hi all, Anyone got any suggestions for a lightweight server distro for an old motherboard? I've got one of the VIA mini-ITX boards, SP13000, and want to whack something light onto it. It will be working as a file/media server and will be headless, hence will be fiddled via ssh. Obviously there are the usual suspects, debian, centos, but does anyone have any recommendations viv a vis a stripped down distro, sort of like Lubuntu is to Ubuntu? Any thoughts greatly appreciated, Andrew Well, if you are only going to need it as an NAS, why not try FreeNAS? OK, its kernel is BSD rather than Linux, but that shouldn't be a problem. Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Slot war on my system
On Sun, 03 Jun 2012 00:27:09 -0500 Michael Sullivan msulli1...@gmail.com wrote: x11-libs/qt-qt3support:4 (x11-libs/qt-qt3support-4.8.1::gentoo, installed) pulled in by ~x11-libs/qt-qt3support-4.8.1[accessibility=,aqua=,c++0x=,debug=,qpa=] required by (x11-libs/qt-declarative-4.8.1::gentoo, installed) (and 1 more with the same problem) (x11-libs/qt-qt3support-4.7.4::gentoo, ebuild scheduled for merge) pulled in by =x11-libs/qt-qt3support-4.6.3:4[accessibility,kde] required by (net-libs/libbluedevil-1.9::gentoo, installed) =x11-libs/qt-qt3support-4.6.3:4[accessibility,kde] required by (net-wireless/bluedevil-1.1-r2::gentoo, installed) When did you last sync your portage tree? bluedevil-1.1-r2 was removed from the main tree on 2011-10-31 and libbluedevil-1.9 on 2012-02-24. Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Re: xfce4 4.10 xsane and Application -- Run Program not working
On Wed, 30 May 2012 07:12:23 -0600 Joseph syscon...@gmail.com wrote: No, xterm nano doesn't work either. More information than doesn't work either please. For example: Does xterm alone (or whatever terminal you use) start? Does nano start in tty? Does your Xsession error log record anything when 'xterm nano doesn't work'? Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Re: Understanding new ruby dependencies
On Tue, 22 May 2012 23:35:21 -0700
Chris Stankevitz chrisstankev...@gmail.com wrote:
On Tue, May 22, 2012 at 10:13 PM, kwk...@hkbn.net wrote:
I suggest keeping an eye on ${PORTDIR}/profiles/desc directory too.
This is where every one of the USE_EXPAND variables is explained in
details.
Thank you for all your patient help. I've been using Gentoo for years
and for some reason this RUBY thing has me flustered.
1. What on my system is insisting on make.conf RUBY 1.9 USE_EXPAND
changes? An emerge --tree is not giving me a clear answer (as it
usually does). The original post in this thread provides a pastebin
link to back up this claim.
Basically the newslot upgrade ruby 1.8.x - 1.9.x.
For example, you can see that in
${PORTDIR}/dev-ruby/json/json-1.5.4-r1.ebuild there is the line
PDEPEND=
rdoc? ( =dev-ruby/rdoc-3.9.4[ruby_targets_ruby19] )
xemacs? ( app-xemacs/ruby-modes )
Previously in json-1.5.4.ebuild there is no such check, as you can
diff for yourself.
2. If the answer to (1) is the gentoo system itself, then why
doesn't the gentoo system itself update the USE_EXPAND by adding a
reference to ruby19? It appears the gentoo system itself presently
only enables the ruby18 USE_EXPAND.
base $ find /usr/portage/profiles/ | xargs grep RUBY_TARGETS=
/usr/portage/profiles/base/make.defaults:RUBY_TARGETS=ruby18
That is usual... profile changes lag behind the ebuild changes.
4. I run a stable system that is somehow insisting on ruby19. This
webpage http://www.gentoo.org/proj/en/prog_lang/ruby/index.xml says
ruby19 is not for use on production systems. Why the disconnect?
Perhaps the ruby page is just out of date.
I suppose ruby19 is in a state similar to python3 --- not ready to be
default since *something* break, but it has been out long enough to be
considered stable.
5. I have no idea what RUBY is and have never installed it directly.
Yet I have to understand RUBY USE_EXPANDs which seem to be described
only in the RUBY installation guide, gentoo dev manual, or in ebuild
scripting guides. I am a gentoo layperson in general and am
completely clueless about RUBY in particular. I believe talk about
this required and automatically installed package should appear not
in obscure dev documentation, but in the handbook. Perhaps with more
time/volunteers this would have happened.
TBH, I'm not a ruby person either. The only thing here on my system
that pulls in ruby is dev-texlive/texlive-pictures, plus I need ruby
for some random scripts I pulled from the web (which I could have
rewritten in python or bash but I can't be bothered).
6. Why does emerge insist on me adding USE=ruby_targets_ruby19 to a
bunch of projects, yet the users of this group recommend a change in
make.conf? I suspect the disconnect that the two approaches are
equivalent, just emerge does not have the street smarts to recommend
the proper change.
That is how ebuild (and hence portage) works --- it didn't check
RUBY_TARGETS but instead the specified use flags for dependencies
specified the ebuild. Hence the error message is add use flag bar to
package foo, regardless of whether bar is actually an expanded flag. As
you see in the example above json ebuild tells portage to check
dev-ruby/rdoc is built with ruby_targets_ruby19 use flag (which is what
RUBY_TARGETS=ruby19 would have expanded to) enabled and so that is what
portage did (and screamed when it can't).
Kerwin.
signature.asc
Description: PGP signature
Re: [gentoo-user] Re: [entirely ON topic] Thanks to the devs
On Thu, 24 May 2012 02:03:07 +0200 Alan McKinnon alan.mckin...@gmail.com wrote: On Wed, 23 May 2012 16:42:15 -0700 walt w41...@gmail.com wrote: On 05/23/2012 12:10 PM, Michael Mol wrote: I've noticed a lot of overt participation on this list by gentoo devs[1], especially in scenarios where they're explaining rationales and reasoning behind changes that affect users. I'd just like to say thank you for the work you guys do. Thank you for the work you guys do. I posted the same, but a long time ago, so I'll add a mee too. Another me too. (Thanking the suse devs here would of course be very inappropriate ;) But then you'd be excluding gregkh and *that* would be very inappropriate ;-) Is gregkh still a SUSE dev? I thought he now works for Linux Foundation (since February). Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Re: Understanding new ruby dependencies
On Tue, 22 May 2012 11:01:45 -0700
Chris Stankevitz chrisstankev...@gmail.com wrote:
Apparently I have to add some ruby_targets_ruby19 USE flags to my
system.
No! Don't do that! Instead, you should add a line
RUBY_TARGETS=ruby19
in your make.conf (or RUBY_TARGETS=ruby18 ruby19) and let portage do
the USE_EXPAND to ruby_targets_ruby19 (respectively,
ruby_targets_ruby18 ruby_targets_ruby19) itself for the relevant
packages. See ${PORTDIR}/profiles/desc/ruby_targets.desc for
description. It is much easier and more intuitive this way, since you
are not doing weird things like building package A for ruby18 but not
package B.
Question 2: Does it seem weird that portage wants me to add USE flags
to enable some feature in a package I never heard of and have no
interest in and to top it off has a very weird name
(ruby_targets_ruby19)?
The weird name is a result of USE_EXPANDing RUBY_TARGETS, just like
LINGUAS and SANE_BACKENDS, for example.
Kerwin.
signature.asc
Description: PGP signature
Re: [gentoo-user] Re: Understanding new ruby dependencies
On Tue, 22 May 2012 23:16:00 -0400
Michael Orlitzky mich...@orlitzky.com wrote:
On 05/22/2012 09:10 PM, Chris Stankevitz wrote:
How was I supposed to learn the proper way of dealing with this
RUBY-related system change?
That change was committed two-and-a-half years ago in the eclass:
http://archives.gentoo.org/gentoo-dev/msg_2305dbeaaf5b02cb74a84c9b06333708.xml
and the Gentoo Ruby project has a section on it
http://www.gentoo.org/proj/en/prog_lang/ruby/index.xml
Actually I learnt the RUBY_TARGETS from flameeyes's blog two years ago
http://blog.flameeyes.eu/2010/02/ruby-ng-package-in-a-bottle-or-learn-how-to-write-a-new-ruby-ebuild
The way I learned was by watching the emerge output:
$ emerge -pv dev-ruby/rails
These are the packages that would be merged, in order:
Calculating dependencies... done!
[ebuild N ] app-admin/eselect-rails-0.16 2 kB
[ebuild N ] dev-ruby/tmail-1.2.7.1-r2 USE=-debug -doc -test
RUBY_TARGETS=ruby18 -jruby -ree18 436 kB
and wondering, RUBY_TARGETS, what the hell is that?
I'm sure there's a proper way, but that's the way I've discovered all
of the USE_EXPAND variables. LINGUAS was the first, then I noticed
ALSA_CARDS, APACHE2_MODULES, XFCE_PLUGINS...
The default list can be found in,
/usr/portage/profiles/base/make.defaults
I suggest keeping an eye on ${PORTDIR}/profiles/desc directory too.
This is where every one of the USE_EXPAND variables is explained in
details.
Kerwin.
signature.asc
Description: PGP signature
Re: [gentoo-user] fsck separate /usr
On Sat, 12 May 2012 19:54:24 -0500 Canek Peláez Valdés can...@gmail.com wrote: On Sat, May 12, 2012 at 7:43 PM, Alex Schuster wo...@wonkology.org wrote: Hi there! I'm using the new udev with a separate /usr partition. How do you create your initramfs? The new udev (= 182, I believe) requires the use of an initramfs if you have a separated /usr. It was encrypted, and it seems there is no solution yet for this. dracut has two modules, crypt and crypt-gpg, that maybe do what you are needing. so I moved it over to an unencrypted volume - no problem, /usr is one partition where encryption does not make that much sense anyway. Works, but after an unclean shutdown (reading files in /proc/pid/ was not a good idea) /usr wants to be fsck'ed. But it is already mounted at that stage. That's the reason you need an initramfs. No, that's the reason you want the filesystem's fsck to be included in the initramfs. The boot process just continues, but I wonder what one should do to make the fsck run. Except for using a live cd. With an initramfs. Using initramfs is necessary but itself not sufficient. One can create an initramfs (from scratch) that does nothing but mount /usr (with only busybox and a few /dev nodes, plus whatever other tools needed to find /usr, viz. lvm, cryptsetup and friends, assuming the necessary drivers are built in the kernel and not as modules --- see e.g. the old gentoo wiki at http://www.gentoo-wiki.info/HOWTO_Custom_Initramfs_From_Scratch ). The initramfs needs to have the relevant fsck tools (plus dependencies) if it was to perform fsck. Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] printer
On Mon, 30 Apr 2012 10:04:10 +0200 Stephane Guedon steph...@einstein.22decembre.eu wrote: Le lundi 30 avril 2012 12:59:10 kwk...@hkbn.net a écrit : On Sun, 29 Apr 2012 21:55:08 +0200 Stephane Guedon steph...@einstein.22decembre.eu wrote: Hi everyone I am now forced to replace my epson printer. Anyone think of a printer for which ink is quite cheap (contrary to the epson) and that allow to have status not only in windows ? Epson as an utility to have ink status in windows and linux, but I have my printer on a server and was in hope some could have ink status iin cups ... So, anyone that have a suggestion is welcome ! Is it just me? There is a bad GPG signature (maybe it is due to the accent character(s)). Anyway, IMO hplip is much nicer than mtink (there is the command-line tool hp-level included that displays ink level), and HP ink don't dry up if you leave the printer idle for 2 weeks, unlike Epson. Of course, if you print a lot you should consider laser printer. Kerwin Anyway, I don't print so much, but I need one ! I don't know why you say I have a bad gpg. I think I put it on the gpg servers, so It should be ok. Did you try to search for it ? Thanks I got the key 0x0403A28B2D8DE8FB from zimmerman.mayfirst.org (one of the keys.gnupg.net keyserver) but signature verification fails for both messages. Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] printer
On Sun, 29 Apr 2012 21:55:08 +0200 Stephane Guedon steph...@einstein.22decembre.eu wrote: Hi everyone I am now forced to replace my epson printer. Anyone think of a printer for which ink is quite cheap (contrary to the epson) and that allow to have status not only in windows ? Epson as an utility to have ink status in windows and linux, but I have my printer on a server and was in hope some could have ink status iin cups ... So, anyone that have a suggestion is welcome ! Is it just me? There is a bad GPG signature (maybe it is due to the accent character(s)). Anyway, IMO hplip is much nicer than mtink (there is the command-line tool hp-level included that displays ink level), and HP ink don't dry up if you leave the printer idle for 2 weeks, unlike Epson. Of course, if you print a lot you should consider laser printer. Kerwin signature.asc Description: PGP signature
Re: [gentoo-user] printer
On Mon, 30 Apr 2012 12:59:10 +0800 kwk...@hkbn.net wrote: there is the command-line tool hp-level included that displays ink level Oops, should be hp-levels, not hp-level. Kerwin signature.asc Description: PGP signature
Re: [gentoo-user] sys-libs/glibc-2.15-r1
On Mon, 23 Apr 2012 14:42:28 +0200 Volker Armin Hemmann volkerar...@googlemail.com wrote: Strange, never had problems in that regard. See, for example, the problem described in thread http://archives.gentoo.org/gentoo-user/msg_264b8690e0ab67e3f55c0967cba101ec.xml from this list last November with glibc-2.14 upgrade, which I also encountered when upgrading my stable amd64. Maybe glibc-2.15 has improved, but I certainly won't risk it when the time comes. I'd rather spend (waste?) time to buy that peace of mind, rather than being locked out of my box. Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Not able to login as root after update
On Mon, 23 Apr 2012 07:17:45 +0530 Nilesh Govindrajan cont...@nileshgr.com wrote: Hi, There's some weird problem that has cropped up after update yesterday. I'm not able to login as root on the tty nor as su -l and neither using sudo su -l. Thankfully I had sudo, so I can get root by sudo -i, or else it would've been a complete disaster. The packages which were upgraded yesterday which can be causing this is pambase, shadow and glibc. Anybody has idea how to fix this? (PS: I know very well that logging in as root is a bad practice and I never do that, I just use su -l, so no flames please). If you mean shadow-4.1.5-r1 and pambase-20120417, read the previous pambase/shadow warning thread from 3 days ago, or bug #412721, or you can read it on phajdan.jr's blog http://phajdan-jr.blogspot.com/2012/04/watch-out-for-shadow-415-r1-and-pambase.html Anyway, if you upgrade glibc, you must recompile pam afterwards manually (sorry it wasn't clear if you recompiled pam after glibc upgrade) since it uses dlopen() which isn't catched by revdep-rebuild. Safest is to recompile the toolchain, then system, then world, but that could take a long time. Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] sys-libs/glibc-2.15-r1
On Mon, 23 Apr 2012 09:12:48 +0700 Pandu Poluan pa...@poluan.info wrote: On Apr 23, 2012 1:09 AM, Volker Armin Hemmann volkerar...@googlemail.com wrote: Am Sonntag, 22. April 2012, 19:52:16 schrieb Stefan G. Weichinger: Just browsed the changelog of glibc-2.15: http://sourceware.org/ml/libc-alpha/2012-03/msg00836.html When I read the NEWS section there with all that optimized stuff I wonder if it makes any sense to rebuild packages here after upgrading glibc? no, because it is a library. You make use of it anyway. What about statically linked packages? Rgds, Bless those who keeps on telling people there is no need to rebuild packages after glibc upgrade, for they must have not used pam or any other packages that uses dlopen(). Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] Another plan for /usr and udev-181
On Sat, 14 Apr 2012 07:35:52 +0100 Neil Bothwick n...@digimed.co.uk wrote: On Fri, 13 Apr 2012 18:44:37 -0700, fe...@crowfix.com wrote: What annoys me the most about this forced change is that I like the old unix style of a single minimal base partition for booting, and being able to manage all the other partitions while unmounted in single user mode. In my case, /usr is an LVM partition precisely because I want to sit in single user mode while resizing it (it seems to keep on growing ...). It's been safe to increase the size of mounted filesystems for years. But if you can enlarge /usr while using it, you can do the same for/. So if that's your only reason for a separate /usr... / on LVM is officially not supported (in the sense there are no official documentation about it) in Gentoo, and is discouraged in the Gentoo LVM installation guide. Has been the case since the beginning, although there are unofficial wiki and mailinglist/forum posts about it. Of course, / on LVM would require an initrd. That's one reason why many of us using LVM keeps /usr on LVM while / as a physical partition. This allows for maximum flexibility, and is a supported legacy config without an initrd. I may add many of us had bad experience with initrd from binary distros rendering system unbootable (I've been there with Debian and Arch --- back in 2003 or so you cannot uninstall currently running kernel initrd after installing a new kernel, or else the next time your newly installed kernel won't boot. Also sometimes the newly installed kernel+initrd won't boot, and neither would the old kernel+initrd...). Of course, now that separate /usr requires an initrd, one might as well put / on LVM and let busybox in initrd handles the case when LVM goes wrong (urgh!). Still, Gentoo doesn't officially support this configuration. -- Kerwin signature.asc Description: PGP signature
Re: [gentoo-user] About ready to move /usr, /var and /home to LVM.
On Sat, 14 Apr 2012 05:32:01 -0500 Dale rdalek1...@gmail.com wrote: Gregory Shearman wrote: In linux.gentoo.user, Dale wrote: I have ran into a issue here. I copied everything over to sdb, my temp drive. When I try to boot it, it still boots from sda which is the primary drive. I can not get it to boot from the copy. I did update the fstab file to point to the new sdb partitions, I use labels for that and they have different names. I also edited grub and told it root was sdb2. When I boot, everything mounted is sda. Did you actually install grub onto your MBR by either: # grub-install --no-floppy /dev/sdb or # grub grub root (hd1,0) grub setup (hd1) grub quit - You didn't actually write down these steps. Are you assuming that we know you've done that? In the past, I never had to install grub to sdb. As long as grub is installed to one drive, I can boot a OS from any drive. I did this when I used to have Mandrake and Gentoo installed. I had Mandrake installed on sda and Gentoo on sdb. I only had one /boot partition which was on sda1. It had the kernel for both Mandrake and Gentoo in it and sda1 was used for both. So, has something changed that if I want to boot from a second drive I have to install grub to its MBR first? When the BIOS finishes and loads grub, doesn't it always load from the first drive? If that is true, doesn't it ignore the MBR on the second drive? It can't load both MBRs right? Yes, if you want to boot from another drive, that drive needs to have a usable MBR (or GPT equivalent). The BIOS (or UEFI) dictates which MBR to load first, and GRUB doesn't come into it until BIOS found it and loaded it. This is usually done in the boot sequence config option in BIOS, although it can be temporarily overridden at boot time by pressing a suitable key. This isn't making sense. I have done this many times in the past with no problems but now something is different. I need help figuring out what. There are many ways this can go wrong. Most probably BIOS boot loading sequence has changed (e.g. if you plug in a USB stick and save boot sequence where the USB stick is tried first, then what happened when you remove the stick and reboot is anybody's guess, because the BIOS will try to outsmart you in guessing what that invalid first boot device should have been). Or maybe you had /dev/sdb disk as the first boot disk all along, the previous absence of a bootloader means BIOS tried the next one silently... My own safety net is to have /dev/sda1 and /dev/sdb1 pretty much the same, except the grub.conf has a difference of a useless title line to indicate which disk it was. Dale :-) :-) Kerwin. signature.asc Description: PGP signature
Re: [gentoo-user] About ready to move /usr, /var and /home to LVM.
On Sat, 14 Apr 2012 06:52:20 -0500 Dale rdalek1...@gmail.com wrote: Well, I installed grub to the second drives MBR. I even changed the BIOS to see that drive as the main or first drive. It still boots the old drive. I looked in dmesg and saw where it is supposed to point to the tmp drive and it still boots the old drive even tho it is told not to. Let's see, boot a CD, just do a reinstall from scratch and call it a day. This is ridiculous when you can't tell a boot loader to boot the second drive and it actually do it. Heaven forbid if I had two Linux OSs on here. :-) :-) It sounds like GRUB made the MBR on /dev/sdb to use /dev/sda1 as its root, so maybe something like # grub --no-floppy grub find /boot/grub/stage1 (hd0,0) (hd1,0) Then making GRUB install on /dev/sda pointing to /dev/sda1 grub device (hd0) /dev/sda grub root (hd0,0) grub setup (hd0) and now install on /dev/sdb pointing to /dev/sdb1 grub device (hd0) /dev/sdb grub root (hd0,0) grub setup (hd0) Then you can quit GRUB by issuing grub quit The point being that once you put in the line device (hd0) /dev/sdb, GRUB will *think* that (hd0) refers to the disk /dev/sdb, so the next command root (hd0,0) just means the first partition on this disk will serve as /boot, rather than (hd1,0) which points to 1st partition on the *other* disk, which is possibly where GRUB got confused. Kerwin. signature.asc Description: PGP signature
