Re: blacklist doesn't blacklist !!
On Tue, Mar 03, 2009 at 05:44:11PM -0300, macdowell@dpf.gov.br wrote: > Hi > > > I put ssd in my blacklist but after boot there it comes again. > > m...@hp:~$ cat /etc/modprobe.d/blacklist | grep ssb > blacklist ssb > m...@hp:~$ lsmod | grep ssb > ssb43140 0 > pcmcia 38680 1 ssb > pcmcia_core41508 2 ssb,pcmcia > m...@hp:~$ > > > Why ? Perhaps it's loaded during boot in the initrd. Try rebuild your initrd with : # update-initramfs -u signature.asc Description: Digital signature
Re: install Etch with raid level 10
Le Wednesday 26 September 2007 18:58:14 A J Stiles, vous avez écrit : [...] > Also, *don't* use a RAID1 for swap space: it impacts performance with > little practical benefit. Use separate swap partitions from each drive > (just to keep the partitioning schemes the same) instead. Sorry, but you'd better have your swap on a mirror ! With a different swap on each disk, if you loose one disk with active swap on it, you will surely crash the system. The memory on it will be lost. For the performance problem, I even use LVM for my swap ! If there is swapping, it's not normal., and performance will be bad anyway. I prefer not to swap, and to have a secured, not so speedy swap. signature.asc Description: This is a digitally signed message part.
Re: install Etch with raid level 10
Le Wednesday 26 September 2007 18:55:50 Lennart Sorensen, vous avez écrit : > On Wed, Sep 26, 2007 at 06:53:16PM +0200, Gilles Mocellin wrote: > > Instead of having md over md device, I suggest to do stripped LVM over > > RAID 1. And it is supported during installation ! > > But if you ever add more devices to the LVM you loose the stripe feature > on those new additions (as far as I understand the LVM documentation). As you can't choose on wich device your LV's PE goes, you have only 2 choice to add space. - If you want to add LVs, just make another VG with 2 new RAID 1 PVs. - If you wan't to resize actual LVs on a full VG, you have to grow the RAIDs. It's hard but I have donne it once, all online. On a SAN, it's easy to resize a LUN, but if you use disks, you'll have to replace the disks with bigger ones. > No idea if you can resize raid0 or raid1. You must use the --grow option of mdadm, and there's chances you'll have to synchronise your RAID each time one RAID is grown. If you can't replace your disks an only can add more disks, a backup/restore will be required to recreate the stripped LVs across more RAID1 PVs. Perhaps pvmove can help to switch to another VG without the backup/restore phase. signature.asc Description: This is a digitally signed message part.
Re: install Etch with raid level 10
Le Wednesday 26 September 2007 16:48:14 Scott Lair, vous avez écrit : > Dave Ewart wrote: [...] > > Normally, one makes a RAID-10 software setup as follows: > > > > mdadm --create /dev/md0 --level=1 --raid-devices=2 /dev/sd[ab]1 > > mdadm --create /dev/md1 --level=1 --raid-devices=2 /dev/sd[cd]1 > > mdadm --create /dev/md2 --level=0 --raid-devices=2 /dev/md[01] [...] > I've got some time to mess with it for a while and I'll post > back the results. > > thanks, > > scott Instead of having md over md device, I suggest to do stripped LVM over RAID 1. And it is supported during installation ! signature.asc Description: This is a digitally signed message part.
Re: Kernel and Xen on an Intel Quad-Core Xeon E5320 processor
Le Tuesday 11 September 2007 04:05:08 GNUbie, vous avez écrit : > Hello Gilles, > [...] > I already installed the linux-image-2.6.18-5-xen-vserver-amd64 on my server > and decided to have a VServer and Xen setup. But, I can't find a good > HOWTO on such setup. Is it having VServer inside Xen or the other way > around? Vserver "in" Xen. In facts vservers in the same Xen virtual machine. > My Debian GNU/Linux Etch AMD64 main system is currently running > the > linux-image-2.6.18-5-xen-vserver-amd64 kernel and Xen is already running: > > # xm list > Name ID Mem(MiB) VCPUs State Time(s) > Domain-0 0 256 8 r-201.4 Here, I see that you have only the Xen Dom0 running. From here you already can run vservers, as your Dom0 kernel is a vserver one. If you only want to have Linux vservers, no other operating systems, You don't need Xen at all ! Use linux-image-*-vserver-amd64 kernel. You you really want to mix Xen and Vservers, What I would do myself, is a light Xen Dom0 without the vserver patch, and a DomU specialized for vservers, with the vserver and xen kernel. Other DomU for other virtual machines, perhaps Windows one ore full virtuallized other distro... By not using vservers directly on Dom0, you'll keep the possibility of online migration of the Xen DomU. But, I must say, I nerver used vservers... specialy with Xen... signature.asc Description: This is a digitally signed message part.
Re: Kernel and Xen on an Intel Quad-Core Xeon E5320 processor
Le Tuesday 04 September 2007 16:42:54 GNUbie, vous avez écrit : > Hello Gilles, > > Sorry for the late response to your e-mail. After some vacation, I more late than you ! > linux-image-2.6.18-5-xen-amd64 > linux-image-2.6.18-5-xen-vserver-amd64 This question ahas been answered many times on several lists, but perhaps not on this. So : Vserver is another virtualisation technologie (sort of enhanced chroot). It is not in the standard kernel, so you have à linux-image*vserver package. But, as it is not like Xen (para or full virtualization), it can be used with it. You can Have a server with several virtual machines with Xen, and in on of them (Dom0 or DomU), you can have some vservers if you use a linux-image-*-xen-vserver-*. > Also, what is the difference between xen-linux-system-2.6.18-5-xen-amd64 > and the xen-linux-system-2.6.18-5-xen-vserver-amd64 and which of these two > shall I install? Install xen-linux-system-2.6.18-5-xen-amd64. > Please advice. > > Thank you in advance. > > GNUbie signature.asc Description: This is a digitally signed message part.
Re: Kernel and Xen on an Intel Quad-Core Xeon E5320 processor
Le Tuesday 28 August 2007 08:05:59 GNUbie, vous avez écrit : > Hello Jim, > > On 8/28/07, Jim Crilly <[EMAIL PROTECTED]> wrote: > > You can't give each VM (or domU in Xen terms) all of your memory so at > > the very least you'll have to rethink that part of your setup. > > Then how will I allocate memory for each domU? What's the best practice > for this kind of setup? > > As for the CPUs, > > > Xen 3.0 and up does seem to support SMP domUs but I can't imagine it > > would be a very good idea to give multiple domUs all 4 CPUs. > > What can you advice then based from your experience? > > If you have this kind of machine, how will you design your Xen domUs in > such a way that you can properly utilize all your hardware components with > the optimum performance for your network services as well as hosting > different web domains you plan to deploy in a production environment? > > Please advice. > > Thank you in advance. > > GNUbie Start simple : - 1 vCPU per domU, try to let one for dom0, specially if it manages a complex storage (soft RAID, DRBD, LVM, iSCSI...) - Let at least 512Mo for dom 0, spread the rest on your domUs depending on what they do. Good to know, you can change the memory allocated for a domU online ! Make some tests, load tests. If you know or you see that some tasks are slow because they can't be parallelised, add a vCPU to the domU. The essential here, is to know what your domUs will really do, and how much you planned. signature.asc Description: This is a digitally signed message part.
Re: Help on understanding noapic nolapic noacpi/acpi=off boot options
Le Monday 16 July 2007 14:08:35 Juan P. Rigol Sanchez, vous avez écrit : > Hi all, > > I have installed (after several attempts) debian lenny amd64 on a new HP > Compaq dc7700 pc (intel core 2 duo 6600 with integrated Q963/Q965 > chipset & graphics), but the system hangs if "noapic nolapic noacpi > acpi=off" options are not given at boot time. > I was wondering how these options affect normal operation of the pc > (fan/cooling, peripherals?, performance??) > (kernel 2.6.21-2-amd64 x86_64 ; with 2.6.18-4-amd64 boots using just > acpi=off). > > Thanks in advance. > > Regards, > Juan P. Rigol Just for info, I have deployed such systems. To keep acpi and the CPU cool, don't put noacpi noapic... I found that just one option was necessary : pci=conf1 Don't ask me what it does... maybe to force pci plug and play config one ? signature.asc Description: This is a digitally signed message part.
Re: Toshiba Core 2 Duo
Le Tuesday 10 July 2007 22:39:18 Seb, vous avez écrit : > Hi, > > Is AMD64 the right Debian distribution to install on a Toshiba Core 2 Duo > laptop? It is definitely a 64 bit architecture, but am not sure which > Debian distribution is the correct one for this laptop. Thanks for any > suggestions. Yes, this is the right architecture. Don't choose ia64 which is for Itanium. But, i386 will also work. Be aware that not all programs are ported to a 64bits environment. The most known of them is non-free : flashplayer. There is ways to make it work by the way, but not straight. signature.asc Description: This is a digitally signed message part.
Re: squares in a 32bits gtk application
Le Monday 09 July 2007 16:59:15 José Alburquerque, vous avez écrit : > Jaime Ochoa Malagón wrote: > > Hi list, > > > > I want to try google desktop but I want to try this outside the > > chroot, inside the chroot that shows correctly outside all the > > characters are replaced by squares. > > > > Any hint? > > Thanks > > I see this when I run a gtk application compiled for a 32bit system (one > that I compiled myself some time ago). > > -Jose Yes, with 32bits binaries (on a 64bit system) it does not use the good GTK libraries. Even with ia32-libs-gtk installed. I have the problem with the VMware server console, here's how I launch it : VMWARE_USE_SHIPPED_GTK=yes vmware-server-console But thats's because vmware is aware of it. A general way will be to launch in a 32bit chroot, or to modify LD_LIBRARY_PATH sith the paths to 32bits libraries. signature.asc Description: This is a digitally signed message part.
Re: libstdc++2.10-glibc2.2
Le vendredi 16 mars 2007 17:28, Hans-Ulrich Klein a écrit : > Hi List, > I installed Debian testing (amd64 port) on a HP ProLiant server and it > really works fine. As a last installation step, I want to install some > HP software offering remote control of the server's health status. HP > offers its software as deb-packages for amd64. Unfortunately, the > HP-package depends on "libstdc++2.10-glibc2.2". The installation guide ( > http://h20293.www2.hp.com/catalog_content/prod/product/install_page/T2812AA >.html ) says: > > hpasm required library or package: >[...] >ia32-libs (for AMD64 only) >[...] >Note: As for hpasm 7.70 AMD64, you must force an install of i386 > library, libstdc++2.10-glibc2.2, run > #dpkg -i --force-architecture > libstdc++2.10-glibc2.2_2.95.4-22_i386.deb > > Is this really the appropriate ways to install the i386 libs? It seems > to be "grubby" for me. > > Thanks in advance, > Hans-Ulrich As it's said, you have to force installation of the 32bits version. I had the same problem last week. Download the libstdc++2.10-glibc2.2_2.95.4-22_i386.deb from the 32bits repository and #dpkg -i --force-architecture libstdc++2.10-glibc2.2_2.95.4-22_i386.deb Unfortunately, there is no way to build a repository for amd64 with 386 packages. We can hope that when etch is stable, HP will give good packages, not depending on such things... pgpMC87hCqilF.pgp Description: PGP signature
Re: 32 bit application
Le mardi 13 mars 2007 21:56, Gudjon I. Gudjonsson a écrit : > Thanks Lennart for the answer > > Þann Þriðjudagur 13 mars 2007 20:14 skrifaði Lennart Sorensen: > > On Tue, Mar 13, 2007 at 05:18:23PM +0100, Gudjon I. Gudjonsson wrote: [...] > > > (/usr1/bin/pjob-manager-real:11161): GdkPixbuf-WARNING **: Error > > > loading XPM image loader: Unable to load image-loading > > > module: /usr/lib/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: > > > /usr/lib/gtk-2.0/2.4.0/loaders/libpixbufloader-xpm.so: cannot open > > > shared object file: No such file or directory > > > > > > (/usr1/bin/pjob-manager-real:11161): libpjm-WARNING **: fail to setup > > > icon: pjm-license ((null)) > > > > > > I get more or less the same results if I try to install realplayer into > > > the 64 bit environment. Does anyone know the solution. > > > > > > The program should most probably be run in chroot but there is a reason > > > why it is done like that. > > > > 32bit programs need 32bit libraries. It appears you are missing some of > > those. Since it can be tricky to have both 32 and 64bit libraries > > installed at the same time (at least they have to be in different > > locations since the filenames are the same) the chroot is much simpler, > > since there you can run en entire debian installation and manage > > upgrades and such using apt and friends, rather than having to manually > > update specific files for 32bit use. > > > > -- > > Len Sorensen > > I know that chroot is much more convenient and I use it myself most often. > In this case I don't and it is a long history that I am not going to tell. > The problem is the same as when I tried (for theoretical purposes :) to > install realplayer in the 64 bit part of my computer. So if somebody has > managed to make realplayer work in a 64 bit environment you may tell me. > > Regards > Gudjon You can install 32bits libs. For your case, specially the package ia32-libs-gtk. pgpWZuxN6Jztl.pgp Description: PGP signature