[gentoo-user] maintaining clones
Hi, I have 4 identical machines, they only differ in the 2 files /etc/conf.d/hostname /etc/conf.d/net I'd like to maintain only one of them (updating GenToo upto several times a week) and 'rsync' the other ones. Now, rsync'ing a life root filesystem is risky. I don't see any problems for the FS holding /usr. The question is, which directories (files) are effected by an 'emerge' operation and which of these do need rsyncing? E.g. /var/tmp doesn't need to be synchronized while (probably) /var/lib does. Many thanks for your help, Helmut. -- Helmut Jarausch Lehrstuhl fuer Numerische Mathematik RWTH - Aachen University D 52056 Aachen, Germany
Re: [gentoo-user] maintaining clones
On 31 Jul, Neil Bothwick wrote: On Fri, 31 Jul 2009 09:07:14 +0200 (CEST), Helmut Jarausch wrote: I have 4 identical machines, they only differ in the 2 files /etc/conf.d/hostname /etc/conf.d/net I'd like to maintain only one of them (updating GenToo upto several times a week) and 'rsync' the other ones. Now, rsync'ing a life root filesystem is risky. I don't see any problems for the FS holding /usr. The whole idea sounds a little risky. I'd use binary packages to keep the other machines up to date. Set FEATURES=buildpkg in make.conf on each computer and set PKGDIR to a directory accessible by all over NFS. Run your normal emerge -u --whatever world on the first then run the same with -k on the others. That way they all get the same updates but only the first has to compile them. I'd also set up distcc to reduce compile times, but that's a separate step. Thanks for your help! I've done so in the past but I've made bad experience. Unfortunately portage isn't so clever, yet. Many times (on the 'clones', as well) I had to block packages before emerge and then unblock again. I even had to unmerge some packages temporarily and emerge them later on again. Sometime I have to patch an ebuild file (temporarily) and so on. And let alone updating baselayout. So, it would be much easier if I could simulate cloning the machines each day. Helmut. -- Helmut Jarausch Lehrstuhl fuer Numerische Mathematik RWTH - Aachen University D 52056 Aachen, Germany
Re: [gentoo-user] maintaining clones
On Fri, 31 Jul 2009 09:07:14 +0200 (CEST), Helmut Jarausch wrote: I have 4 identical machines, they only differ in the 2 files /etc/conf.d/hostname /etc/conf.d/net I'd like to maintain only one of them (updating GenToo upto several times a week) and 'rsync' the other ones. Now, rsync'ing a life root filesystem is risky. I don't see any problems for the FS holding /usr. The whole idea sounds a little risky. I'd use binary packages to keep the other machines up to date. Set FEATURES=buildpkg in make.conf on each computer and set PKGDIR to a directory accessible by all over NFS. Run your normal emerge -u --whatever world on the first then run the same with -k on the others. That way they all get the same updates but only the first has to compile them. I'd also set up distcc to reduce compile times, but that's a separate step. -- Neil Bothwick Very funny Scotty.. now beam down my pants! signature.asc Description: PGP signature
Re: [gentoo-user] maintaining clones
On Fri, 31 Jul 2009 09:59:02 +0200 (CEST), Helmut Jarausch wrote: I've done so in the past but I've made bad experience. Unfortunately portage isn't so clever, yet. portage is a lot cleverer than it used to be, especially with regard to blocks. Many times (on the 'clones', as well) I had to block packages before emerge and then unblock again. I even had to unmerge some packages temporarily and emerge them later on again. It sounds like you were using -K rather than -k or trying to build binary packages with --buildpkgonly, which is doomed to failure when trying to build dependent packages. -- Neil Bothwick Maybe... How much are you bribing me this time? signature.asc Description: PGP signature
Re: [gentoo-user] maintaining clones
On 31 Jul, Neil Bothwick wrote: On Fri, 31 Jul 2009 09:59:02 +0200 (CEST), Helmut Jarausch wrote: I've done so in the past but I've made bad experience. Unfortunately portage isn't so clever, yet. portage is a lot cleverer than it used to be, especially with regard to blocks. Yes, I'm using portage-2.2_rc33 but still ... Many times (on the 'clones', as well) I had to block packages before emerge and then unblock again. I even had to unmerge some packages temporarily and emerge them later on again. It sounds like you were using -K rather than -k or trying to build binary packages with --buildpkgonly, which is doomed to failure when trying to build dependent packages. No, on the 'master' machine I have in /etc/make.conf FEATURES=buildpkg -stricter and on the clones I use '-k'. But still, portage checks dependencies on the clone. And, if it sees a problem (like those when upgrading to kde-4 or qt-4.5.x) it refused to merge the binary package. Helmut. -- Helmut Jarausch Lehrstuhl fuer Numerische Mathematik RWTH - Aachen University D 52056 Aachen, Germany
Re: [gentoo-user] maintaining clones
On Fri, 31 Jul 2009 10:46:17 +0200 (CEST), Helmut Jarausch wrote: No, on the 'master' machine I have in /etc/make.conf FEATURES=buildpkg -stricter and on the clones I use '-k'. But still, portage checks dependencies on the clone. And, if it sees a problem (like those when upgrading to kde-4 or qt-4.5.x) it refused to merge the binary package. Those sort of blocks, such as when switching from monolithic to split ebuilds, are rare and easily dealt with. Dealing with unexpected side effects of syncing in-use files could be a lot more problematic. -- Neil Bothwick All mail what i send is thoughly proof-red, definately! signature.asc Description: PGP signature
[gentoo-user] Re: How send mail when user login on ssh or local ?
On 07/30/2009 06:12 PM, Alan McKinnon wrote: On Friday 31 July 2009 00:05:16 Harry Putnam wrote: Somebody know how I to so send mail with IP and Date/time when same user login on shell ( remote or local ) ? I work with another admin's and I never told me when they access and for what my server to do something, I try log but this can be erased and maybe mail can help me about access and with this I can Ask about this access. Do you really think the other admins would be erasing logs? That's what I was thinking. If you don't trust the other admin, then either: 1. You need to stop being the admin, or 2. The other person needs to stop being the admin. This is not a technical problem, it does not need a technical solution. It is a human problem and it needs a human solution. This sounds nasty. Of course it is nasty - it intended it to be. But it's also true. In 25 years in this game, I have never found the above to be false. Trying to do anything about it is a fool's game and down that path lies madness. I agree 100% with Alan and it is a log-standing policy of mine that I will not share admin responsibilities with someone not on my staff. Until you put that in place however, you might consider setting the system syslog to log to an external system that only you control. The other admin can turn off logging but you will know when he's done so.
Re: [gentoo-user] Changing xorg.conf at runtime (nVidia cards)
Hi, On Fri, Jul 31, 2009 at 11:23, Dalerdalek1...@gmail.com wrote: Mike Mazur wrote: On Fri, Jul 31, 2009 at 05:44, Fernando Antunesfs.antu...@gmail.com wrote: I don't use xorg.conf in my notebook Lenovo T61, Intel 965GSM , xorg and xfce ~x86, gentoo 2.6.30 with KMS, anymore. Both kernel and X switch to 1280x800 resolution automatically, xinerama is disable. Interesting, so you don't have /etc/X11/xorg.conf file at all? If you have hal enabled and xorg-server-1.5 or higher, then you don't have to have a xorg.conf file. This is from what I have read. I have yet to get that hal thing to work here. If you are still on the old xorg-server then you have to have a xorg.conf file. I tried this earlier, running startx without an /etc/X11/xorg.conf in place, but X didn't start. The error message was something along the lines of No screen found. I'm running: - x11-base/xorg-server-1.5.3-r6 - x11-base/xorg-x11-7.2 - sys-apps/hal-0.5.11-r9 Mike
Re: [gentoo-user] {OT} SSD instead of RAID1?
I read on this list that the kernel needs *some* swap, even just a tiny amount, to function properly. Is that true? If so, do you think it would be OK to put this tiny amount of swap on a cheap SSD? I have no swap and things work just fine. (8 gigs of RAM) Obviously, running without swap increases the chances of you running out of memory, but that has never happened to me. I've been setting up all of my systems according to this, creating a 512MB swap partition: http://www.gentoo.org/doc/en/handbook/handbook-amd64.xml?part=1chap=4 If I have 4GB RAM, all I'm accomplishing with swap is increasing this to 4.5GB? If my system requires 4.6GB at some point, I'm in the same position I would be in if I had no swap and 4.1GB requirement? As far as I understand it, correct. What have you guys found to be the minimum RAM necessary for a Gentoo system without swap that doesn't lock up? My laptop has 1GB RAM and 512MB swap, and when I disabled the swap the system locked up during the first big emerge. I guess that means I need somewhere between 1GB and 1.5GB. I'd say it's a pretty normal laptop/workstation. - Grant
Re: [gentoo-user] Changing xorg.conf at runtime (nVidia cards)
On Fri, Jul 31, 2009 at 11:28 AM, Mike Mazur mma...@gmail.com wrote: Hi, On Fri, Jul 31, 2009 at 11:23, Dalerdalek1...@gmail.com wrote: Mike Mazur wrote: On Fri, Jul 31, 2009 at 05:44, Fernando Antunesfs.antu...@gmail.com wrote: I don't use xorg.conf in my notebook Lenovo T61, Intel 965GSM , xorg and xfce ~x86, gentoo 2.6.30 with KMS, anymore. Both kernel and X switch to 1280x800 resolution automatically, xinerama is disable. Interesting, so you don't have /etc/X11/xorg.conf file at all? If you have hal enabled and xorg-server-1.5 or higher, then you don't have to have a xorg.conf file. This is from what I have read. I have yet to get that hal thing to work here. If you are still on the old xorg-server then you have to have a xorg.conf file. I tried this earlier, running startx without an /etc/X11/xorg.conf in place, but X didn't start. The error message was something along the lines of No screen found. I'm running: - x11-base/xorg-server-1.5.3-r6 - x11-base/xorg-x11-7.2 - sys-apps/hal-0.5.11-r9 Iḿ running : xorg-server : 1.6.2-r1 x11 : 7.4 hal : 0.5.13-r2 kernel : 2.6.30-gentoo-r4 x11-video-intel : 2.8.0 Mike
Re: [gentoo-user] {OT} SSD instead of RAID1?
Anyway, the point of all this is to prevent an HD failure from stopping the system. An SSD is much safer, right? SSDs are still relatively new technology, so predicting failure rates is less reliable. What's wrong with using RAID-1? It's proven technology and totally resistant to a single HD failure. This was Grant's original question - whether SSD / flash technology is more reliable than RAID-1 of conventional disks? - and one to which no-one appeared comfortable giving a categorical answer. Stroller. I've come up with a couple reasons to wait a bit longer to switch my important systems to SSD. 1. SLC is faster and (more importantly) should last much longer than MLC. The Super Talent Ultradrive 32GB drives are priced ~$120 for MLC and ~$350 for SLC, so I'd like to wait for that SLC price to drop. It's worth mentioning though, that even conservative estimates of MLC lifetimes put them far beyond that of HD drives. 2. SSD fIrmware is being updated relatively frequently right now (especially newer SSDs) and all data is lost during a firmware update. I'm sold on SSDs as RAID1 replacements though. BTW, I read that Samsung manufactures the memory for all major brand SSDs (including Super Talent). - Grant
[gentoo-user] Cannot emerge libgksu: lacks XML::Parser
I've got portage's dev-perl/XML-Parser-2.36, but nevertheless several packages have started to fail during emerge for lack of Perl's XML::Parser. Even the simple script fails to run under perl: #!/usr/bin/perl -w use XML::Parser; print yes\n; Do I have to get it from CPAN? Consider libgksu; here's the end of the emerge output, where XML::Parser is required by intltool: checking for perl... /usr/bin/perl checking for XML::Parser... configure: error: XML::Parser perl module is required for intltool !!! Please attach the following file when seeking support: !!! /var/tmp/portage/x11-libs/libgksu-2.0.9/work/libgksu-2.0.9/config.log * * ERROR: x11-libs/libgksu-2.0.9 failed. * Call stack: * ebuild.sh, line 49: Called src_compile * environment, line 2896: Called gnome2_src_compile * environment, line 2264: Called gnome2_src_configure * environment, line 2278: Called econf '--enable-nls' '--disable-gtk-doc' * ebuild.sh, line 534: Called die * The specific snippet of code: * die econf failed * The die message: * econf failed * * If you need support, post the topmost build error, and the call stack if relevant. * A complete build log is located at '/var/tmp/portage/x11-libs/libgksu-2.0.9/temp/build.log'. * The ebuild environment file is located at '/var/tmp/portage/x11-libs/libgksu-2.0.9/temp/environment'. * Failed to emerge x11-libs/libgksu-2.0.9, Log file: '/var/tmp/portage/x11-libs/libgksu-2.0.9/temp/build.log' -- Kevin O'Gorman, PhD
Re: [gentoo-user] maintaining clones
On Fri, 31 Jul 2009 09:07:14 +0200 (CEST) Helmut Jarausch jarau...@igpm.rwth-aachen.de wrote: Hi, I have 4 identical machines, they only differ in the 2 files /etc/conf.d/hostname /etc/conf.d/net I'd like to maintain only one of them (updating GenToo upto several times a week) and 'rsync' the other ones. hmm... network booting? network mounting? install packages once on one system, share them with everyone. Share passwd/shadow files and the like manually, or symlink them to skeletal versions symlinked to somewhere that can be obscured and replaced by a network boot. you could even boot them from thumb drives or cds. of course, it would be a good bit of work to configure initially, and might not go whithout a hitch.
Re: [gentoo-user] maintaining clones
On Fri, 31 Jul 2009 09:57:48 +0100 Neil Bothwick n...@digimed.co.uk wrote: Dealing with unexpected side effects of syncing in-use files could be a lot more problematic. perhaps a digest of some kind? md5 the files, write up a little script to keep the rest of the nodes synced up?
[gentoo-user] Re: Cannot emerge libgksu: lacks XML::Parser
Kevin O'Gorman wrote:
I've got portage's dev-perl/XML-Parser-2.36, but nevertheless several
packages have started to fail during emerge for lack of Perl's
XML::Parser.
Even the simple script fails to run under perl:
#!/usr/bin/perl -w
use XML::Parser;
print yes\n;
Do I have to get it from CPAN?
Consider libgksu; here's the end of the emerge output, where
XML::Parser is required by intltool:
checking for perl... /usr/bin/perl
checking for XML::Parser... configure: error: XML::Parser perl module
is required for intltool
!!! Please attach the following file when seeking support:
!!! /var/tmp/portage/x11-libs/libgksu-2.0.9/work/libgksu-2.0.9/config.log
*
* ERROR: x11-libs/libgksu-2.0.9 failed.
* Call stack:
* ebuild.sh, line 49: Called src_compile
* environment, line 2896: Called gnome2_src_compile
* environment, line 2264: Called gnome2_src_configure
* environment, line 2278: Called econf '--enable-nls'
'--disable-gtk-doc'
* ebuild.sh, line 534: Called die
* The specific snippet of code:
* die econf failed
* The die message:
* econf failed
*
* If you need support, post the topmost build error, and the call
stack if relevant.
* A complete build log is located at
'/var/tmp/portage/x11-libs/libgksu-2.0.9/temp/build.log'.
* The ebuild environment file is located at
'/var/tmp/portage/x11-libs/libgksu-2.0.9/temp/environment'.
*
Failed to emerge x11-libs/libgksu-2.0.9, Log file:
'/var/tmp/portage/x11-libs/libgksu-2.0.9/temp/build.log'
Have you recently changed the ithreads USE flag on dev-lang/perl? If so,
then you will need to remerge all ebuilds that installed files in
/usr/lib*/perl5/vendor_perl/${PERL_VER}/${CHOST%%-*}-linux if ithreads was
disabled, and /usr/lib*/perl5/vendor_perl/${PERL_VER}/${CHOST%%-*}-linux-
thread-multi if ithreads was enabled.
Also, if you have recently upgraded perl, the same provisions apply.
NOTE: In the above expansion, the following applies [sorry if you already
know this]:
- lib* is lib on x86, lib64 on amd64
- ${PERL_VER} is your perl version, probably 5.8.8
- ${CHOST%%-*} is the part of the CHOST before the first -:
* on x86, it will be one of i486, i586, or i686
* on amd64, it will be x86_64
If you aren't on x86 or amd64, I assume you can figure it out :).
--
ABCD
