Re: [gentoo-user] Dual or Quad CPU complications?
I've only ever used systems with a single CPU. I'm looking for a new host for a dedicated server (suggestions?) and it looks like I'll probably choose a machine with two or four CPUs. What sort of complications does that add to set up and/or maintenance with Gentoo? none also, forget numa. You won't deal with douzends of cores each using local memory and acccession the memory managed by the other cores. It depends on his application, maybe his application does benefit on NUMA architecture. Until we don't know what he's running, we can't really say this or that architecture/technology is of no use ;) So Volker, what applications are you running (and BTW: what volume of data are you managing, how many users, ...)? This will helps us help you :) Rafa It's an apache2/mysql server with a medium amount of traffic. - Grant
Re: [gentoo-user] Dual or Quad CPU complications?
On Fri, Dec 14, 2012 at 8:43 AM, Grant emailgr...@gmail.com wrote: [...] So if I have 2 physical CPU's with 4 cores each and I enable SMP, I'm using 8 cores? Can NUMA be either enabled or disabled when using more than one physical CPU, or is it required? NUMA is a hardware architecture. It's how you access memory on a hardware level: NUMA = Non Uniform Memory Access vs a UMA architecture of typical (old/legacy) SMP systems (UMA = Uniform Memory Access). In a UMA system, all the memory belongs to all the sockets. In a NUMA system, each socket has it's own local memory. In modern (x86-64) processors, each socket has it's own memory controller so each socket controls its own local memory. If one socket runs out of memory it can ask another socket to lend him some memory. In a UMA system, no socket has to ask since memory is global and belongs to all sockets so if one socket uses up all the memory ... the rest starve. In NUMA, there's more control over who uses what (be it cores or RAM). If you have a modern dual or quad (or higher #) socket system ... you've got NUMA architecture and you can't get rid of it, it's hardware, not software. NUMA is not bad nor good. It's transparent to you. If your SW supports threads, OpenMP, ... you'll be using it without knowing. That doesn't mean you can't tweak performance and use numactl tools, cgroups, ... to increase performance. You can :) HTH Rafa
Re: [gentoo-user] {OT} dedicated server or cloud server?
Am 14.12.2012 08:49, schrieb Grant: Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant Last time I did the calculation, a dedicated or normal virtualized infrastructure was more cost effective as long as you could accurately predict the performance you need. Cloud services only really help if you need a high dynamic range regarding scale and performance, e.g. a service that could get a lot of new users very fast or is only really active for short time spans. Regards, Florian Philipp signature.asc Description: OpenPGP digital signature
Re: [gentoo-user] Dual or Quad CPU complications?
So if I have 2 physical CPU's with 4 cores each and I enable SMP, I'm using 8 cores? Can NUMA be either enabled or disabled when using more than one physical CPU, or is it required? NUMA is a hardware architecture. It's how you access memory on a hardware level: NUMA = Non Uniform Memory Access vs a UMA architecture of typical (old/legacy) SMP systems (UMA = Uniform Memory Access). In a UMA system, all the memory belongs to all the sockets. In a NUMA system, each socket has it's own local memory. In modern (x86-64) processors, each socket has it's own memory controller so each socket controls its own local memory. If one socket runs out of memory it can ask another socket to lend him some memory. In a UMA system, no socket has to ask since memory is global and belongs to all sockets so if one socket uses up all the memory ... the rest starve. In NUMA, there's more control over who uses what (be it cores or RAM). If you have a modern dual or quad (or higher #) socket system ... you've got NUMA architecture and you can't get rid of it, it's hardware, not software. So I must enable CONFIG_NUMA for more than one physical CPU, and disable it for only one physical CPU? - Grant NUMA is not bad nor good. It's transparent to you. If your SW supports threads, OpenMP, ... you'll be using it without knowing. That doesn't mean you can't tweak performance and use numactl tools, cgroups, ... to increase performance. You can :) HTH Rafa
Re: [gentoo-user] {OT} dedicated server or cloud server?
Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant Last time I did the calculation, a dedicated or normal virtualized infrastructure was more cost effective as long as you could accurately predict the performance you need. Cloud services only really help if you need a high dynamic range regarding scale and performance, e.g. a service that could get a lot of new users very fast or is only really active for short time spans. Doesn't a good cloud server also have potentially higher availability compared to dedicated? - Grant
Re: [gentoo-user] Dual or Quad CPU complications?
On Fri, Dec 14, 2012 at 10:44 AM, Grant emailgr...@gmail.com wrote: So if I have 2 physical CPU's with 4 cores each and I enable SMP, I'm using 8 cores? Can NUMA be either enabled or disabled when using more than one physical CPU, or is it required? NUMA is a hardware architecture. It's how you access memory on a hardware level: NUMA = Non Uniform Memory Access vs a UMA architecture of typical (old/legacy) SMP systems (UMA = Uniform Memory Access). In a UMA system, all the memory belongs to all the sockets. In a NUMA system, each socket has it's own local memory. In modern (x86-64) processors, each socket has it's own memory controller so each socket controls its own local memory. If one socket runs out of memory it can ask another socket to lend him some memory. In a UMA system, no socket has to ask since memory is global and belongs to all sockets so if one socket uses up all the memory ... the rest starve. In NUMA, there's more control over who uses what (be it cores or RAM). If you have a modern dual or quad (or higher #) socket system ... you've got NUMA architecture and you can't get rid of it, it's hardware, not software. So I must enable CONFIG_NUMA for more than one physical CPU, and disable it for only one physical CPU? Yup. But ... Why would you want to disable a socket (CPU)? If you disable a socket (CPU) ... you lose the memory attached to that socket (CPU) not to mention you lose those cores ;) A better solution would be to use cgroups or numactl tools to pin a certain process to a set of cores and a memory region. If you really want to deactivate cores (but not the whole socket), you can type: echo 0 /sys/devices/system/cpu/cpu1/online This would deactivate core #1. You can deactivate as many cores as you wish, except for core #0. This can be done without rebooting your server (aka during run time). Your memory will not be affected, but you will have less cores (and theoretically more memory bandwidth). I say theoretically because you always have to benchmark these things with YOUR application (remember logic NEVER applies to real life ;) If you want to check the # of cores you've got: cat /proc/interrupts | grep CPU Other possibilities such as cat /proc/cpuinfo or dmesg, ... can be useful too for this: your choice, FLOSS gives you options. If you want to activate the previously deactivated core, you can run: echo 1 /sys/devices/system/cpu/cpu1/online Now ... be sure your core numbering is the expected core numbering. IOW, not all server vendors follow the same numbering scheme so core #1 in vendor A's server could be core #2 in vendor B's server. Never trust logic ;) As I mentioned previously: test/benchmark YOUR software. DON'T trust logic or generic benchmarks or web pages with results. Trust YOUR results only. HTH Rafa
Re: [gentoo-user] emerge sets syntax (@world vs. world)
On Wed, Dec 12, 2012, at 19:22, Francesco Turco wrote: I'm still not convinced. emerge(1) man page for portage-2.1.11.37 already contains the following command example: emerge --update --newuse --deep @world And: emerge --update @world But not a single example without the at sign. I also found this (old) blog post from Portage developer Zac Medico: http://blogs.gentoo.org/zmedico/2010/09/07/portage_2-1-9_release/. It says: Package set names in emerge arguments have to be prefixed with @ (exceptions: ‘world’ and ‘system’ can be used without the prefix). So it seems that since version 2.1.9 @world and world (and @system and system) are just treated in the same way, but prefixing them with the at symbol is more future-proof. I contacted Zac Medico and he said me: Yes, @world is more future proof. I don't plan to deprecate the old way any time soon, but it could happen at some point in the future. It seems pretty clear to me. Plain world is not wrong, but @world is the way to go.
[gentoo-user] Anyone switched to eudev yet?
Howdy, I noticed eudev has hit the tree. Has anyone used it yet? If so, any issues? Did you just uninstall udev and install eudev in one step or some other way? I'm thinking of switching and getting rid of the init thingy but curious as to what others may have ran into. Thanks much. Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] {OT} dedicated server or cloud server?
On Thu, Dec 13, 2012 at 11:49:23PM -0800, Grant wrote: Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant My data is allergic to the cloud ... too much pixey dust. -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] {OT} dedicated server or cloud server?
On Fri, 14 Dec 2012 06:39:03 -0600 Bruce Hill da...@happypenguincomputers.com wrote: On Thu, Dec 13, 2012 at 11:49:23PM -0800, Grant wrote: Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant My data is allergic to the cloud ... too much pixey dust. If I get up from my desk, walk down the corridor and turn right, I find myself right in the middle of a technical team that built and delivers cloud. I do not use cloud. This is not a co-incidence. -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] {OT} dedicated server or cloud server?
Doesn't a good cloud server also have potentially higher availability compared to dedicated? Perhaps at your price point through redundancy which could be applied to dedicated all be it at higher cost and so potentially still more reliable and certainly more secure and also tested in almost any case (lookup the paper about timing attacks on amazon services etc.). -- ___ 'Write programs that do one thing and do it well. Write programs to work together. Write programs to handle text streams, because that is a universal interface' (Doug McIlroy) ___
[gentoo-user] Re: Anyone switched to eudev yet?
On 14/12/12 14:19, Dale wrote: I'm thinking of switching and getting rid of the init thingy Huh?
Re: [gentoo-user] {OT} dedicated server or cloud server?
On Fri, 14 Dec 2012 02:00:54 -0800 Grant emailgr...@gmail.com wrote: Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant Last time I did the calculation, a dedicated or normal virtualized infrastructure was more cost effective as long as you could accurately predict the performance you need. Cloud services only really help if you need a high dynamic range regarding scale and performance, e.g. a service that could get a lot of new users very fast or is only really active for short time spans. Doesn't a good cloud server also have potentially higher availability compared to dedicated? Potentially? Yes. In reality? No. It's not the virtualization that breaks, it's all the surrounding infrastructure, especially Layer 2. You will not believe how fragile that stuff can get. In the old days, a small slip up could isolate a small part of the network. These days, a small slip-up easily ripples though the entire network and takes down all of it, and sadly this is not rare. The networking needs of VMs are radically different from the traditional, and this is the side-effect: fragility. -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] Re: Anyone switched to eudev yet?
Nikos Chantziaras wrote: On 14/12/12 14:19, Dale wrote: I'm thinking of switching and getting rid of the init thingy Huh Right now, I have /usr on a separate partition so I would need a init thingy to boot. When I switch to eudev, that won't be required, from what I have read anyway. I didn't want the init thingy to begin with either. Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] {OT} dedicated server or cloud server?
On Fri, Dec 14, 2012 at 03:19:58PM +0200, Alan McKinnon wrote: On Fri, 14 Dec 2012 06:39:03 -0600 Bruce Hill da...@happypenguincomputers.com wrote: My data is allergic to the cloud ... too much pixey dust. If I get up from my desk, walk down the corridor and turn right, I find myself right in the middle of a technical team that built and delivers cloud. I do not use cloud. This is not a co-incidence. We have reached common ground...thanks for the note. Or as another once said, Great minds think alike. ;) -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] {OT} dedicated server or cloud server?
On Fri, Dec 14, 2012 at 03:24:03PM +0200, Alan McKinnon wrote: Doesn't a good cloud server also have potentially higher availability compared to dedicated? Potentially? Yes. In reality? No. It's not the virtualization that breaks, it's all the surrounding infrastructure, especially Layer 2. You will not believe how fragile that stuff can get. In the old days, a small slip up could isolate a small part of the network. These days, a small slip-up easily ripples though the entire network and takes down all of it, and sadly this is not rare. The networking needs of VMs are radically different from the traditional, and this is the side-effect: fragility. And it happens *every* day, all over the place. Yesterday I went to Coburn's Supply to get a box of air filters for our HVAC units. They couldn't get in. One guy says, I'm on the internet, don't know what's wrong. The other guy says, It's _them_ ... So they couldn't even tell me the price of the filters, or if they had them. So one guy walks in the back and does it the old fashioned way -- he just looks. (Reminds me a bit of sneakernet.) By the time he gets back, and tries again, Hey, it's working now ... quick! Welcome to the cloud. Your packet has reached critical mass. Please reboot. -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 07:41:45AM -0600, Dale wrote: Right now, I have /usr on a separate partition so I would need a init thingy to boot. When I switch to eudev, that won't be required, from what I have read anyway. I didn't want the init thingy to begin with either. Dale Let me translate his Mississippi English... define: init thingy initrd: http://en.wikipedia.org/wiki/Initrd My file server has /boot with / on /dev/md0 (well, see here:) mingdao@server ~ $ egrep -v (^#|^ *$) /etc/fstab /dev/md0/ xfs inode64,logbsize=262144 0 1 /var/swapfile1 swapswapdefaults 0 0 /dev/system/var /varxfs defaults 0 0 /dev/system/usr /usrxfs defaults 0 0 /dev/system/home/home xfs defaults 0 0 /dev/storage/photos /photos xfs users,rw 0 0 /dev/storage/backups/backupsxfs users,rw 0 0 /dev/storage/offload/offloadntfsdefaults 0 0 /dev/storage/peter /peter xfs defaults 0 0 /dev/storage/jeremiah /jeremiah xfs defaults 0 0 And no init thingy anywhere on this LAN. ;) -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] Re: Anyone switched to eudev yet?
Bruce Hill wrote: On Fri, Dec 14, 2012 at 07:41:45AM -0600, Dale wrote: Right now, I have /usr on a separate partition so I would need a init thingy to boot. When I switch to eudev, that won't be required, from what I have read anyway. I didn't want the init thingy to begin with either. Dale Let me translate his Mississippi English... define: init thingy initrd: http://en.wikipedia.org/wiki/Initrd My file server has /boot with / on /dev/md0 (well, see here:) mingdao@server ~ $ egrep -v (^#|^ *$) /etc/fstab /dev/md0 / xfs inode64,logbsize=262144 0 1 /var/swapfile1swapswapdefaults 0 0 /dev/system/var /varxfs defaults 0 0 /dev/system/usr /usrxfs defaults 0 0 /dev/system/home /home xfs defaults 0 0 /dev/storage/photos /photos xfs users,rw 0 0 /dev/storage/backups /backupsxfs users,rw 0 0 /dev/storage/offload /offloadntfsdefaults 0 0 /dev/storage/peter/peter xfs defaults 0 0 /dev/storage/jeremiah /jeremiah xfs defaults 0 0 And no init thingy anywhere on this LAN. ;) Pretty much yea. I started making a init thing when they were talking about not supporting /usr on a separate partition. Then about a month ago eudev was announced which means we can boot with /usr on a separate partition and no init thingy, like it used to be. My basic question is this, has anyone started using eudev yet? From my understanding it is basically udev with the files where they used to be before they changed things. I'm thinking about switching but wondering what all is involved. It appears to be as simple as unmerge udev and emerge eudev and restart eudev. Is it really THAT simple? Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 4:19 AM, Dale rdalek1...@gmail.com wrote: Howdy, I noticed eudev has hit the tree. Has anyone used it yet? If so, any issues? Did you just uninstall udev and install eudev in one step or some other way? I'm thinking of switching and getting rid of the init thingy but curious as to what others may have ran into. Thanks much. Dale Even if someone has, and clearly _someone_ out there has or it likely wouldn't even be visible yet, but even if 10 or 20 people have, and even if all of their results are fine because they are high skill set folks, why would that change how you are running your machines? I suspect this is about your (and my) dislike for dealing with initrd on a box at home. Gentoo doesn't make it at all easy so we're in that together. However so what if someone has used it? Let it get used for 6 months. Let it go stable. Why bother with a piece of software that won't really improve your life now that you do have your 'init thingy'? Just my view, Mark
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 08:25:08AM -0600, Dale wrote: Pretty much yea. I started making a init thing when they were talking about not supporting /usr on a separate partition. Then about a month ago eudev was announced which means we can boot with /usr on a separate partition and no init thingy, like it used to be. My basic question is this, has anyone started using eudev yet? From my understanding it is basically udev with the files where they used to be before they changed things. I'm thinking about switching but wondering what all is involved. It appears to be as simple as unmerge udev and emerge eudev and restart eudev. Is it really THAT simple? Dale What Mark wrote you is golden. I might only add that if you put: =sys-fs/udev-181 into /etc/portage/package.mask you will have the present stable udev from *before* those weirdos starting messing it up, forcing systemd to take over udev. And, you won't be required to have an initrd image. (At sys-fs/udev-171-r9 as of Fri Dec 14 09:37:06 CST 2012). And the only USE flag you *need* for that version of udev would be rule_generator. If you don't know that you need another, you probably don't. Bruce -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] Anyone switched to eudev yet?
Mark Knecht wrote: On Fri, Dec 14, 2012 at 4:19 AM, Dale rdalek1...@gmail.com wrote: Howdy, I noticed eudev has hit the tree. Has anyone used it yet? If so, any issues? Did you just uninstall udev and install eudev in one step or some other way? I'm thinking of switching and getting rid of the init thingy but curious as to what others may have ran into. Thanks much. Dale Even if someone has, and clearly _someone_ out there has or it likely wouldn't even be visible yet, but even if 10 or 20 people have, and even if all of their results are fine because they are high skill set folks, why would that change how you are running your machines? I suspect this is about your (and my) dislike for dealing with initrd on a box at home. Gentoo doesn't make it at all easy so we're in that together. However so what if someone has used it? Let it get used for 6 months. Let it go stable. Why bother with a piece of software that won't really improve your life now that you do have your 'init thingy'? Just my view, Mark Well, it appears that one version is stable: root@fireball / # equery list -p eudev * Searching for eudev ... [-P-] [ ] sys-fs/eudev-0:0 [-P-] [ ~] sys-fs/eudev-1_beta1-r1:0 [-P-] [ -] sys-fs/eudev-:0 root@fireball / # The first one is not keyworded or masked. You are right, I don't like the init fix because when I used Mandrake, it caused me all sorts of problems. That and the upgrade process for Mandrake is the reason I switched to Gentoo. If eudev is ready, then so am I. Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 7:48 AM, Dale rdalek1...@gmail.com wrote: Mark Knecht wrote: On Fri, Dec 14, 2012 at 4:19 AM, Dale rdalek1...@gmail.com wrote: Howdy, I noticed eudev has hit the tree. Has anyone used it yet? If so, any issues? Did you just uninstall udev and install eudev in one step or some other way? I'm thinking of switching and getting rid of the init thingy but curious as to what others may have ran into. Thanks much. Dale Even if someone has, and clearly _someone_ out there has or it likely wouldn't even be visible yet, but even if 10 or 20 people have, and even if all of their results are fine because they are high skill set folks, why would that change how you are running your machines? I suspect this is about your (and my) dislike for dealing with initrd on a box at home. Gentoo doesn't make it at all easy so we're in that together. However so what if someone has used it? Let it get used for 6 months. Let it go stable. Why bother with a piece of software that won't really improve your life now that you do have your 'init thingy'? Just my view, Mark Well, it appears that one version is stable: root@fireball / # equery list -p eudev * Searching for eudev ... [-P-] [ ] sys-fs/eudev-0:0 [-P-] [ ~] sys-fs/eudev-1_beta1-r1:0 [-P-] [ -] sys-fs/eudev-:0 root@fireball / # The first one is not keyworded or masked. You are right, I don't like the init fix because when I used Mandrake, it caused me all sorts of problems. That and the upgrade process for Mandrake is the reason I switched to Gentoo. If eudev is ready, then so am I. Dale Well, OK, so if you want to call version 0.0 stable then I guess that meets the rules of portage anyway. However version '0.0' doesn't sound like anything that's seen the light of day, been used by lots of people and proven robust and stable. At least to me it sounds like a place holder... This is just my view, but it goes something like this: 1) Unless someone tells me why a really new package helps me then I go slow, most especially if it could have a large impact like a new version of udev might. 2) Somewhere in the install guide, or elsewhere, I don't remember, it says something like 'don't expect ~packages to work correctly. We do what we can to check them but you should expect things to break'. And then most importantly, again from memory and paraphrased 'If you don't know how to fix things when they do break don't use ~packages'. I let these few sentences guide a lot of my Gentoo maintenance here at home. I mask packages (good info from Bruce about which to mask) and wait for the heavy lifters to shake things out a bit before I update things that might take more than 5 minutes to fix. Again, all my systems are stable with ~amd64 only when required to get an app, but that's just me. Good luck with whatever path you take. Cheers, Mark
Re: [gentoo-user] Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 09:48:12AM -0600, Dale wrote: Well, it appears that one version is stable: root@fireball / # equery list -p eudev * Searching for eudev ... [-P-] [ ] sys-fs/eudev-0:0 [-P-] [ ~] sys-fs/eudev-1_beta1-r1:0 [-P-] [ -] sys-fs/eudev-:0 root@fireball / # The first one is not keyworded or masked. You are right, I don't like the init fix because when I used Mandrake, it caused me all sorts of problems. That and the upgrade process for Mandrake is the reason I switched to Gentoo. If eudev is ready, then so am I. Dale Kindly read /usr/portage/sys-fs/eudev/eudev-0.ebuild especially: pkg_pretend() { einfo Please note that sys-fs/eudev-0 is actually a mirror of sys-fs/udev-171-r9, einfo which is necessary to handle the package name change when migrating to eudev einfo from udev 180 -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
[gentoo-user] ~amd64 compatibility with modern cpus
Something went haywire with my 8 or 9 year old dual Opteron ~amd64 system last night. I may have a bricked system. I haven't given up yet, but I may have to buy a replacement system. I have external USB drive backups, but the only other computer I have right now is an old Mac laptop which can't read Linux LVM partitions. Questions: 1. I don't remember, and can't look up, the make.conf processor flags I emerge with. But it is dual Opterons, and ~amd64. How compatible could that be with modern Intel CPUs? I know Intel adopted the extra registers of the AMD64 instruction set, but are there other differences which would prevent an Opteron system from running as is under an Intel processor? Maybe AMD still sells Opterons, and I will be stuck with building a system. 2. Is it feasible to buy some commodity box, like from Dell, with an Intel processor, and plug in my two SATA SSD drives and get a console boot? I don't give a fig right now about any GUI interface, and even Internet is not the problem. If it will boot and run emerges, I can import the source files for X and Ethernet and other peripherals via USB stick. But SATA drivers ... 3. My kernels always have just about every driver compiled in as modules, an old habit from when I used to swap in PCI cards like crazy. I don't remember now how many SATA drivers are built in and how many are modules; if the commodity box needs SATA drivers which aren't built in, that could get tricky. Are there boot command line options to preload certain modules? Might not do me any good. I think I could scrape by with USB modules, but not SATA. For the curious, here is wat happened. When I left off last night, the USB keyboard was only recognized when I unplugged all other USB devices, and the system hung at the grub point, with a blank screen. A reboot failed because it couldn't find the root=/dev/sde drive. But the USB keyboard was working because I used it in grub to select a new 3.7.0 kernel (had been running 3.6.8). A second reboot ignored the USB keyboard and generated an ATA error I had never seen before for every ATA drive and some I don't have, all the way up to ATA13 before I rebooted it again. I haven't got it to boot even this far since, so I can't regenerate that error. There was a 5 second or so delay between these errors, making me think the ATAnn designator might not be different drives, just retries. It booted a rescue DVD, but without the keyboard it was kind of pointless, and it hung after showing two lines which I believe are unrelated other than a place marker (generating xxx key, generating RSA key). The keyboard wasn't even recognized by the BIOS. I finally disconnected every USB device, all the ubs, and then the keyboard worked. But when I left it last night, it wouldn't even bring up the grub screen. All the BIOS screens show the usual disk drives. The system was working perfectly fine before all hell broke loose. The keyboard was recognized during grub the first time, but after that only if all other USB devices were disconnected. The disk drives acted funny during the boot, first with the unknown root- device error, then with the funky ATA errors, and finally with not even bringing up grub. I will try some more desperate tricks today, like reconnecting the USB pile to see if it at least boots the disks again - is my choice between disks and keyboard? I will find out. My best guess right now is that booting 3.7.0 is what clobbered things; whether I added a option which loaded bad firmware, or 3.7.0 is broken, I have no idea. It could well be something unrelated to 3.7.0. My goal for today is to try to get keyboard and disk working, then boot with 3.6.8. -- Felix Finch, a la mode
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On 2012-12-14 10:39 AM, Bruce Hill da...@happypenguincomputers.com wrote: What Mark wrote you is golden. I might only add that if you put: =sys-fs/udev-181 into /etc/portage/package.mask you will have the present stable udev from*before* those weirdos starting messing it up, forcing systemd to take over udev. Hmmm... For some reason I have masked my udev at 171... Are you saying I can change the mask to 181 and not have to worry about having an unbootable system due to my /usr being on a separate partition? Thanks
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, 14 Dec 2012 15:26:25 +0200 Nikos Chantziaras rea...@gmail.com wrote: On 14/12/12 14:19, Dale wrote: I'm thinking of switching and getting rid of the init thingy Huh? Once upon a time, not so long ago, Dale happened to try and make an initrd. He followed the rules and the docs and it blew up in his face. It blew up more in his face than HAL did even longer ago (and that was pretty spectacular). Somewhere in that thread the words init thingy were used in frustration, and it's since become a meme, a disparaging reference to initrd for those who don't like the idea of them. Dale is getting famous at this, if ever we need someone to find the really bad bugs in basic software, just mark it stable enough so that Dale will use it. He will find the bugs that no-one else ever could, it's just a knack he has. Long live Dale. :-) -- Alan McKinnon alan.mckin...@gmail.com
Re: [gentoo-user] ~amd64 compatibility with modern cpus
(Admittedly quick and dirty response) On Fri, Dec 14, 2012 at 11:18 AM, fe...@crowfix.com wrote: Something went haywire with my 8 or 9 year old dual Opteron ~amd64 system last night. I may have a bricked system. I haven't given up yet, but I may have to buy a replacement system. I have external USB drive backups, but the only other computer I have right now is an old Mac laptop which can't read Linux LVM partitions. Questions: 1. I don't remember, and can't look up, the make.conf processor flags I emerge with. But it is dual Opterons, and ~amd64. How compatible could that be with modern Intel CPUs? I know Intel adopted the extra registers of the AMD64 instruction set, but are there other differences which would prevent an Opteron system from running as is under an Intel processor? Maybe AMD still sells Opterons, and I will be stuck with building a system. I'm not aware of any currently-sold AMD or Intel x86 processors (except, *maybe* Atom) which don't handily support x86_64. You should have no problem here. 2. Is it feasible to buy some commodity box, like from Dell, with an Intel processor, and plug in my two SATA SSD drives and get a console boot? I don't give a fig right now about any GUI interface, and even Internet is not the problem. If it will boot and run emerges, I can import the source files for X and Ethernet and other peripherals via USB stick. But SATA drivers ... Depends on if you have the necessary drivers installed, but yes, it's a recoverable scenario. You might need to boot a live CD and compile and install a kernel with the right drivers. It'd be like installing Gentoo fresh, but skipping 95% of the handbook. 3. My kernels always have just about every driver compiled in as modules, an old habit from when I used to swap in PCI cards like crazy. I don't remember now how many SATA drivers are built in and how many are modules; if the commodity box needs SATA drivers which aren't built in, that could get tricky. Are there boot command line options to preload certain modules? Might not do me any good. I think I could scrape by with USB modules, but not SATA. NAFAIK. I'd just use the Gentoo live CD image. For the curious, here is wat happened. When I left off last night, the USB keyboard was only recognized when I unplugged all other USB devices, and the system hung at the grub point, with a blank screen. A reboot failed because it couldn't find the root=/dev/sde drive. But the USB keyboard was working because I used it in grub to select a new 3.7.0 kernel (had been running 3.6.8). A second reboot ignored the USB keyboard and generated an ATA error I had never seen before for every ATA drive and some I don't have, all the way up to ATA13 before I rebooted it again. I haven't got it to boot even this far since, so I can't regenerate that error. There was a 5 second or so delay between these errors, making me think the ATAnn designator might not be different drives, just retries. It booted a rescue DVD, but without the keyboard it was kind of pointless, and it hung after showing two lines which I believe are unrelated other than a place marker (generating xxx key, generating RSA key). The keyboard wasn't even recognized by the BIOS. I finally disconnected every USB device, all the ubs, and then the keyboard worked. But when I left it last night, it wouldn't even bring up the grub screen. All the BIOS screens show the usual disk drives. The system was working perfectly fine before all hell broke loose. The keyboard was recognized during grub the first time, but after that only if all other USB devices were disconnected. The disk drives acted funny during the boot, first with the unknown root- device error, then with the funky ATA errors, and finally with not even bringing up grub. I will try some more desperate tricks today, like reconnecting the USB pile to see if it at least boots the disks again - is my choice between disks and keyboard? I will find out. My best guess right now is that booting 3.7.0 is what clobbered things; whether I added a option which loaded bad firmware, or 3.7.0 is broken, I have no idea. It could well be something unrelated to 3.7.0. My goal for today is to try to get keyboard and disk working, then boot with 3.6.8. Pull out an old PS2 keyboard. Sometimes, that's the easiest way to get things going. -- :wq
Re: [gentoo-user] Anyone switched to eudev yet?
Mark Knecht wrote: On Fri, Dec 14, 2012 at 7:48 AM, Dale rdalek1...@gmail.com wrote: Mark Knecht wrote: On Fri, Dec 14, 2012 at 4:19 AM, Dale rdalek1...@gmail.com wrote: Howdy, I noticed eudev has hit the tree. Has anyone used it yet? If so, any issues? Did you just uninstall udev and install eudev in one step or some other way? I'm thinking of switching and getting rid of the init thingy but curious as to what others may have ran into. Thanks much. Dale Even if someone has, and clearly _someone_ out there has or it likely wouldn't even be visible yet, but even if 10 or 20 people have, and even if all of their results are fine because they are high skill set folks, why would that change how you are running your machines? I suspect this is about your (and my) dislike for dealing with initrd on a box at home. Gentoo doesn't make it at all easy so we're in that together. However so what if someone has used it? Let it get used for 6 months. Let it go stable. Why bother with a piece of software that won't really improve your life now that you do have your 'init thingy'? Just my view, Mark Well, it appears that one version is stable: root@fireball / # equery list -p eudev * Searching for eudev ... [-P-] [ ] sys-fs/eudev-0:0 [-P-] [ ~] sys-fs/eudev-1_beta1-r1:0 [-P-] [ -] sys-fs/eudev-:0 root@fireball / # The first one is not keyworded or masked. You are right, I don't like the init fix because when I used Mandrake, it caused me all sorts of problems. That and the upgrade process for Mandrake is the reason I switched to Gentoo. If eudev is ready, then so am I. Dale Well, OK, so if you want to call version 0.0 stable then I guess that meets the rules of portage anyway. However version '0.0' doesn't sound like anything that's seen the light of day, been used by lots of people and proven robust and stable. At least to me it sounds like a place holder... This is just my view, but it goes something like this: 1) Unless someone tells me why a really new package helps me then I go slow, most especially if it could have a large impact like a new version of udev might. 2) Somewhere in the install guide, or elsewhere, I don't remember, it says something like 'don't expect ~packages to work correctly. We do what we can to check them but you should expect things to break'. And then most importantly, again from memory and paraphrased 'If you don't know how to fix things when they do break don't use ~packages'. I let these few sentences guide a lot of my Gentoo maintenance here at home. I mask packages (good info from Bruce about which to mask) and wait for the heavy lifters to shake things out a bit before I update things that might take more than 5 minutes to fix. Again, all my systems are stable with ~amd64 only when required to get an app, but that's just me. Good luck with whatever path you take. Cheers, Mark That's all true, hence my question. I'm not sure I want to use the very first version so I thought it worth asking first. Since it is a fork, one could think it would be safe enough but then again, it is the very first one. It is stable according to that but is it really? That is why I asked. It seems no one has used it yet tho since no one has fessed up to installing it, other than testers and such I guess. ;-) Guess I'll wait a bit and see what else changes. Current udev is working for the moment at least. I do plan to abandon udev as soon as I think it is safe tho. Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Anyone switched to eudev yet?
Bruce Hill wrote: On Fri, Dec 14, 2012 at 09:48:12AM -0600, Dale wrote: Well, it appears that one version is stable: root@fireball / # equery list -p eudev * Searching for eudev ... [-P-] [ ] sys-fs/eudev-0:0 [-P-] [ ~] sys-fs/eudev-1_beta1-r1:0 [-P-] [ -] sys-fs/eudev-:0 root@fireball / # The first one is not keyworded or masked. You are right, I don't like the init fix because when I used Mandrake, it caused me all sorts of problems. That and the upgrade process for Mandrake is the reason I switched to Gentoo. If eudev is ready, then so am I. Dale Kindly read /usr/portage/sys-fs/eudev/eudev-0.ebuild especially: pkg_pretend() { einfo Please note that sys-fs/eudev-0 is actually a mirror of sys-fs/udev-171-r9, einfo which is necessary to handle the package name change when migrating to eudev einfo from udev 180 A, so it needs a little more time yet then. I can do that, for a little while longer. Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 8:23 AM, Alan McKinnon alan.mckin...@gmail.com wrote: SNIP Long live Dale. +1
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 11:23 AM, Alan McKinnon alan.mckin...@gmail.com wrote: On Fri, 14 Dec 2012 15:26:25 +0200 Nikos Chantziaras rea...@gmail.com wrote: On 14/12/12 14:19, Dale wrote: I'm thinking of switching and getting rid of the init thingy Huh? Once upon a time, not so long ago, Dale happened to try and make an initrd. He followed the rules and the docs and it blew up in his face. It blew up more in his face than HAL did even longer ago (and that was pretty spectacular). Somewhere in that thread the words init thingy were used in frustration, and it's since become a meme, a disparaging reference to initrd for those who don't like the idea of them. Dale is getting famous at this, if ever we need someone to find the really bad bugs in basic software, just mark it stable enough so that Dale will use it. He will find the bugs that no-one else ever could, it's just a knack he has. Long live Dale. Hear! Hear! Dale, would you be interested in jobs doing software testing? If I see openings, I'd be happy to forward them your way... -- :wq
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 11:28:41AM -0500, Michael Mol wrote: (Admittedly quick and dirty response) Much appreciated. Gives me some hope ... Pull out an old PS2 keyboard. Sometimes, that's the easiest way to get things going. I thought of that -- don't have any. They all got recycled a few years back :-( -- Felix Finch, a la mode
Re: [gentoo-user] {OT} dedicated server or cloud server?
Am 14.12.2012 11:00, schrieb Grant: Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant Last time I did the calculation, a dedicated or normal virtualized infrastructure was more cost effective as long as you could accurately predict the performance you need. Cloud services only really help if you need a high dynamic range regarding scale and performance, e.g. a service that could get a lot of new users very fast or is only really active for short time spans. Doesn't a good cloud server also have potentially higher availability compared to dedicated? - Grant I'd be grateful if anyone can point me at a well conducted study on that topic. Until then I just say that my anecdotal evidence shows the opposite: My cheap-ass virtual server has an uptime of 492 days with only minor, previously announced network outages. During the same time, Amazon EC2 had what, 3 or 4 major outages? Regards, Florian Philipp signature.asc Description: OpenPGP digital signature
Re: [gentoo-user] ~amd64 compatibility with modern cpus
Am 14.12.2012 17:18, schrieb fe...@crowfix.com: Something went haywire with my 8 or 9 year old dual Opteron ~amd64 system last night. I may have a bricked system. I haven't given up yet, but I may have to buy a replacement system. I have external USB drive backups, but the only other computer I have right now is an old Mac laptop which can't read Linux LVM partitions. Questions: 1. I don't remember, and can't look up, the make.conf processor flags I emerge with. But it is dual Opterons, and ~amd64. How compatible could that be with modern Intel CPUs? I know Intel adopted the extra registers of the AMD64 instruction set, but are there other differences which would prevent an Opteron system from running as is under an Intel processor? Maybe AMD still sells Opterons, and I will be stuck with building a system. I guess your Opterons used -march=k8. Except of 3dnow, this should be compatible. You might be lucky. 2. Is it feasible to buy some commodity box, like from Dell, with an Intel processor, and plug in my two SATA SSD drives and get a console boot? I don't give a fig right now about any GUI interface, and even Internet is not the problem. If it will boot and run emerges, I can import the source files for X and Ethernet and other peripherals via USB stick. But SATA drivers ... Yep, SATA drivers will be the biggest issue. Hope you had and will have AHCI. 3. My kernels always have just about every driver compiled in as modules, an old habit from when I used to swap in PCI cards like crazy. I don't remember now how many SATA drivers are built in and how many are modules; if the commodity box needs SATA drivers which aren't built in, that could get tricky. Are there boot command line options to preload certain modules? Might not do me any good. I think I could scrape by with USB modules, but not SATA. Not possible. You need an initrd or a new kernel. How about compiling a new kernel on a different box and using a memory stick for grub + /boot? For the curious, here is wat happened. When I left off last night, the USB keyboard was only recognized when I unplugged all other USB devices, and the system hung at the grub point, with a blank screen. A reboot failed because it couldn't find the root=/dev/sde drive. But the USB keyboard was working because I used it in grub to select a new 3.7.0 kernel (had been running 3.6.8). A second reboot ignored the USB keyboard and generated an ATA error I had never seen before for every ATA drive and some I don't have, all the way up to ATA13 before I rebooted it again. I haven't got it to boot even this far since, so I can't regenerate that error. There was a 5 second or so delay between these errors, making me think the ATAnn designator might not be different drives, just retries. [...] Could be your south bridge. If you want to keep the system, try a different board. Regards, Florian Philipp signature.asc Description: OpenPGP digital signature
Re: [gentoo-user] Re: Anyone switched to eudev yet?
Alan McKinnon wrote: On Fri, 14 Dec 2012 15:26:25 +0200 Nikos Chantziaras rea...@gmail.com wrote: On 14/12/12 14:19, Dale wrote: I'm thinking of switching and getting rid of the init thingy Huh? Once upon a time, not so long ago, Dale happened to try and make an initrd. He followed the rules and the docs and it blew up in his face. It blew up more in his face than HAL did even longer ago (and that was pretty spectacular). Somewhere in that thread the words init thingy were used in frustration, and it's since become a meme, a disparaging reference to initrd for those who don't like the idea of them. Dale is getting famous at this, if ever we need someone to find the really bad bugs in basic software, just mark it stable enough so that Dale will use it. He will find the bugs that no-one else ever could, it's just a knack he has. Long live Dale. :-) You would mention hal wouldn't you? :-@ :-@ I have the flu, nasty one at that, and I really don't need to add hal to my list right now. That said, the Doctor called and the blood tests said I was healthy as a horse, other than being sick as a dog. :/ Sort of like software, supposed to work great out of the box but usually doesn't. Imagine me on windoze. O_O Dale :-) :-) P. S. I'm not sure about the living part yet. It's a theory that is yet to be proven. -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 8:29 AM, Dale rdalek1...@gmail.com wrote: That's all true, hence my question. I'm not sure I want to use the very first version so I thought it worth asking first. Since it is a fork, one could think it would be safe enough but then again, it is the very first one. It is stable according to that but is it really? But portage isn't telling you to use it. It's a decision you're taking in reaction to how your machine is configured, correct? I think the reason I'm not understanding even asking the question is that portage hasn't _told_ you that you need to use it. I may well be wrong but aren't you a stable user? Or are you running bleeding edge? It's likely to be 6 months and probably far longer before there's any push from portage to tell us switch to eudev, if ever. I guess the other question that's lurking here for me is why do you have /usr on a separate partition? What's the usage model that drives a person to do that? The most I've ever done is move /usr/portage and /usr/src to other places. My /usr never has all that much in it beyond those two directories, along with maybe /usr/share. Would it not be easier for you in the long run to move /usr back to / and not have to deal with this question at all? Just my 2 questioning cents, Mark
Re: [gentoo-user] Re: Anyone switched to eudev yet?
Michael Mol wrote: On Fri, Dec 14, 2012 at 11:23 AM, Alan McKinnon alan.mckin...@gmail.com wrote: On Fri, 14 Dec 2012 15:26:25 +0200 Nikos Chantziaras rea...@gmail.com wrote: On 14/12/12 14:19, Dale wrote: I'm thinking of switching and getting rid of the init thingy Huh? Once upon a time, not so long ago, Dale happened to try and make an initrd. He followed the rules and the docs and it blew up in his face. It blew up more in his face than HAL did even longer ago (and that was pretty spectacular). Somewhere in that thread the words init thingy were used in frustration, and it's since become a meme, a disparaging reference to initrd for those who don't like the idea of them. Dale is getting famous at this, if ever we need someone to find the really bad bugs in basic software, just mark it stable enough so that Dale will use it. He will find the bugs that no-one else ever could, it's just a knack he has. Long live Dale. Hear! Hear! Dale, would you be interested in jobs doing software testing? If I see openings, I'd be happy to forward them your way... -- :wq I thought I did that already. ROFL Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Re: {OT} Will ARM take over the world?
On Thu, Dec 13, 2012 at 07:29:21PM +0100, Volker Armin Hemmann wrote SLAX is using KDE4 - and uses 200mb. KDE is flexible. If you have lots of memory, it does use lots of memory. If you don't it doesn't. So don't group it together with 'lets force mono unto our users - for a notes application' gnome or 'you can always add another 4gig' chrome. What about 'lets force dbus/libmpeg2/qt-webkit/phonon and a relational database on our users - for a pdf viewer'? I have a stripped-down no-nonsense install on all my machines with the USE variable beginning with -*, and only necessary stuff being added. When xpdf was deprecated, one of the suggested alternatives was okular. It requires qt3support, which in turn requires the accessibility USE flag. And qt-gui is required which requires the dbus flag. And oh yeah, I'd have to unmask dbus. And dee first dependancy requires dee second dependancy And dee second dependancy requires dee third dependancy And dee third dependancy requires dee fourth dependancy etc, etc, etc After unmasking dbus and adding a bunch of USE flags, I finally got rid of the emerge error messages... USE=accessibility dbus glib qt3support sqlite gstreamer emerge -p okular Total: 59 packages (59 new), Size of downloads: 347,087 kB A stinking pdf viewer requires, amongst other things... libmpeg2 dbus desktop-file-utils strigi xdg-utils qt-webkit phonon and *A RELATIONAL DATABASE* (Hello!?!?), of which the lightest available is sqlite. Don't waste your time trying to convince me that KDE is lightweight. I run ICEWM. See my sig... -- Walter Dnes waltd...@waltdnes.org I don't run desktop environments; I run useful applications
Re: [gentoo-user] Anyone switched to eudev yet?
Mark Knecht wrote: On Fri, Dec 14, 2012 at 8:29 AM, Dale rdalek1...@gmail.com wrote: That's all true, hence my question. I'm not sure I want to use the very first version so I thought it worth asking first. Since it is a fork, one could think it would be safe enough but then again, it is the very first one. It is stable according to that but is it really? But portage isn't telling you to use it. It's a decision you're taking in reaction to how your machine is configured, correct? I think the reason I'm not understanding even asking the question is that portage hasn't _told_ you that you need to use it. I may well be wrong but aren't you a stable user? Or are you running bleeding edge? It's likely to be 6 months and probably far longer before there's any push from portage to tell us switch to eudev, if ever. I guess the other question that's lurking here for me is why do you have /usr on a separate partition? What's the usage model that drives a person to do that? The most I've ever done is move /usr/portage and /usr/src to other places. My /usr never has all that much in it beyond those two directories, along with maybe /usr/share. Would it not be easier for you in the long run to move /usr back to / and not have to deal with this question at all? Just my 2 questioning cents, Mark Portage doesn't tell me to use a lot of things but I do because it makes things easier for me. I don't always wait for portage to tell me something is about to break before I try to switch. It's just like the init thingy. Right now, I'm not required to use it but I do because when the newer udev comes along, I would then be forced too. I didn't want to wait until the last minute then have to scramble around to get my ducks in a row. Sort of learning to prepare better to avoid problems. I have / and /boot on regular partitions but everything else is on LVM including /usr. I got tired of having to move things around for space issues. So, I switched to LVM but not to the point where I needed a init thingy. THAT I wanted to avoid and was until udev threw a wrench into the works. As long as we stick with the current udev I should be fine. If they start moving up to the newer ones tho, I'll be looking into what to do. I'm hoping to switch to eudev and that cure my issues. Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] ~amd64 compatibility with modern cpus
Am Freitag, 14. Dezember 2012, 11:18:21 schrieb fe...@crowfix.com: Something went haywire with my 8 or 9 year old dual Opteron ~amd64 system last night. I may have a bricked system. I haven't given up yet, but I may have to buy a replacement system. I have external USB drive backups, but the only other computer I have right now is an old Mac laptop which can't read Linux LVM partitions. Questions: 1. I don't remember, and can't look up, the make.conf processor flags I emerge with. But it is dual Opterons, and ~amd64. How compatible could that be with modern Intel CPUs? I know Intel adopted the extra registers of the AMD64 instruction set, but are there other differences which would prevent an Opteron system from running as is under an Intel processor? Maybe AMD still sells Opterons, and I will be stuck with building a system. 2. Is it feasible to buy some commodity box, like from Dell, with an Intel processor, and plug in my two SATA SSD drives and get a console boot? I don't give a fig right now about any GUI interface, and even Internet is not the problem. If it will boot and run emerges, I can import the source files for X and Ethernet and other peripherals via USB stick. But SATA drivers ... 3. My kernels always have just about every driver compiled in as modules, an old habit from when I used to swap in PCI cards like crazy. I don't remember now how many SATA drivers are built in and how many are modules; if the commodity box needs SATA drivers which aren't built in, that could get tricky. Are there boot command line options to preload certain modules? Might not do me any good. I think I could scrape by with USB modules, but not SATA. For the curious, here is wat happened. When I left off last night, the USB keyboard was only recognized when I unplugged all other USB devices, and the system hung at the grub point, with a blank screen. A reboot failed because it couldn't find the root=/dev/sde drive. But the USB keyboard was working because I used it in grub to select a new 3.7.0 kernel (had been running 3.6.8). A second reboot ignored the USB keyboard and generated an ATA error I had never seen before for every ATA drive and some I don't have, all the way up to ATA13 before I rebooted it again. I haven't got it to boot even this far since, so I can't regenerate that error. There was a 5 second or so delay between these errors, making me think the ATAnn designator might not be different drives, just retries. It booted a rescue DVD, but without the keyboard it was kind of pointless, and it hung after showing two lines which I believe are unrelated other than a place marker (generating xxx key, generating RSA key). The keyboard wasn't even recognized by the BIOS. I finally disconnected every USB device, all the ubs, and then the keyboard worked. But when I left it last night, it wouldn't even bring up the grub screen. All the BIOS screens show the usual disk drives. The system was working perfectly fine before all hell broke loose. The keyboard was recognized during grub the first time, but after that only if all other USB devices were disconnected. The disk drives acted funny during the boot, first with the unknown root- device error, then with the funky ATA errors, and finally with not even bringing up grub. I will try some more desperate tricks today, like reconnecting the USB pile to see if it at least boots the disks again - is my choice between disks and keyboard? I will find out. My best guess right now is that booting 3.7.0 is what clobbered things; whether I added a option which loaded bad firmware, or 3.7.0 is broken, I have no idea. It could well be something unrelated to 3.7.0. My goal for today is to try to get keyboard and disk working, then boot with 3.6.8. how about a more stable kernel - like 3.4.X? and yes, a confused bios can do a lot of strange things. One thing you might try: disconnect your box from the main for several minutes, reset bios... had to do that dance A LOT with a costly POS asus board... -- #163933
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 11:20:05AM -0500, Tanstaafl wrote: On 2012-12-14 10:39 AM, Bruce Hill da...@happypenguincomputers.com wrote: What Mark wrote you is golden. I might only add that if you put: =sys-fs/udev-181 into /etc/portage/package.mask you will have the present stable udev from*before* those weirdos starting messing it up, forcing systemd to take over udev. Hmmm... For some reason I have masked my udev at 171... Are you saying I can change the mask to 181 and not have to worry about having an unbootable system due to my /usr being on a separate partition? The =sys-fs/udev-181 in /etc/portage/package.mask means that any udev version equal to or greater than udev-181 is masked from the system. Therefore, the latest version under 181 will be used (stable or unstable depending upon /etc/portage/package.accept_keywords). You can see versions/slots easily by: mingdao@workstation ~ $ eshowkw sys-fs/udev Keywords for sys-fs/udev: | | u | | a a p s | n | | l m h i m m p s p | u s | r | p d a p a 6 i p c 3 a x | s l | e | h 6 r p 6 8 p p 6 9 s r 8 | e o | p | a 4 m a 4 k s c 4 0 h c 6 | d t | o --+---+-+--- 141-r1 | ~ ~ ~ + ~ ~ ~ ~ ~ ~ ~ ~ ~ | # 0 | gentoo 146-r1 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | # | gentoo 149| ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | # | gentoo 151-r4 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | # | gentoo 164-r2 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | # | gentoo [I]171-r9 | + + + + + + ~ + + + + + + | o | gentoo 195| ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | # | gentoo 196-r1 | ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ | o | gentoo | o o o o o o o o o o o o o | o | gentoo -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 06:22:10PM +0100, Volker Armin Hemmann wrote: how about a more stable kernel - like 3.4.X? It was running 3.6.8 fine, and ~ kernels for ages before that. The paranoid in me thinks it was 3.7.0, but I really don't know. and yes, a confused bios can do a lot of strange things. One thing you might try: disconnect your box from the main for several minutes, reset bios... had to do that dance A LOT with a costly POS asus board... I unplugged it last night, tried again half an hour later, no joy, so I unplugged it again and have been eating breakfast, got osme errands to run, and then I will try again. What's so frustrating is that the box was working fine, including the keyboard, until that first boot into 3.7.0, where it couldn't find the root drive, and then the keyboard stopped working, even with the BIOS, almost as if 3.7.0 did something nasty and clobbered everything it could get its hands on. Well, I'll try again in a bit. -- Felix Finch, a la mode
[gentoo-user] Question about new USE flags for xf86-video-intel
I'm updating a system that's probably gone 2 or 3 months since its last update. !!! The ebuild selected to satisfy x11-drivers/xf86-video-intel has unmet requirements. - x11-drivers/xf86-video-intel-2.20.13::gentoo USE=dri -glamor -sna -udev -uxa -xvmc The following REQUIRED_USE flag constraints are unsatisfied: any-of ( glamor sna uxa ) This is the first time I've run into these options. A bit of Google detective work turned up some articles... http://www.phoronix.com/scan.php?page=articleitem=intel_glamor_firstnum=1 http://ickle.wordpress.com/2012/06/05/an-evening-with-glamor/ ...which give the impression that SNA is the fastest option overall. The phoronix article also mentioned some instability issues with glamor. I'll probably be going with SNA. Does anybody else have any experience and/or opinions? -- Walter Dnes waltd...@waltdnes.org I don't run desktop environments; I run useful applications
Re: [gentoo-user] Re: Anyone switched to eudev yet?
On Fri, Dec 14, 2012 at 10:52:12AM -0600, Dale wrote: I have the flu, nasty one at that, and I really don't need to add hal to my list right now. That said, the Doctor called and the blood tests said I was healthy as a horse, other than being sick as a dog. :/ Sort of like software, supposed to work great out of the box but usually doesn't. Imagine me on windoze. O_O Dale :-) :-) P. S. I'm not sure about the living part yet. It's a theory that is yet to be proven. Maybe I can give you a run for your money. I hit most corner cases head on. See this bug https://bugs.gentoo.org/show_bug.cgi?id=447184 Today we have major things going on, for which I need my printer to print. And yesterday it stopped printing. It will crash FF when Print is selected, and won't print from Inkscape, LibreOffice, or Evince. There were already problems but now it just rolled over and died. And, no, it's not the printer ... workstation CAN print with this command: lp -d HP_Officejet_Pro_8500_A910 -o scaling=75 Happy-Penguin-Gymnastics/offers/Drop-and-Shop-text and baruch can print to the printer (192.168.100.10:9100) Hill's Law: it works best when you don't need it -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] Question about new USE flags for xf86-video-intel
On Fri, Dec 14, 2012 at 9:44 AM, Walter Dnes waltd...@waltdnes.org wrote: I'm updating a system that's probably gone 2 or 3 months since its last update. !!! The ebuild selected to satisfy x11-drivers/xf86-video-intel has unmet requirements. - x11-drivers/xf86-video-intel-2.20.13::gentoo USE=dri -glamor -sna -udev -uxa -xvmc The following REQUIRED_USE flag constraints are unsatisfied: any-of ( glamor sna uxa ) This is the first time I've run into these options. A bit of Google detective work turned up some articles... http://www.phoronix.com/scan.php?page=articleitem=intel_glamor_firstnum=1 http://ickle.wordpress.com/2012/06/05/an-evening-with-glamor/ ...which give the impression that SNA is the fastest option overall. The phoronix article also mentioned some instability issues with glamor. I'll probably be going with SNA. Does anybody else have any experience and/or opinions? -- Walter Dnes waltd...@waltdnes.org I don't run desktop environments; I run useful applications I have no idea about performance but I've got udev and sna enabled at this time. gandalf ~ # emerge -pv xf86-video-intel These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R] x11-drivers/xf86-video-intel-2.20.13 USE=dri sna udev -glamor -uxa -xvmc 0 kB Total: 1 package (1 reinstall), Size of downloads: 0 kB gandalf ~ # Be forewarned, or please report back if you can, but in my case on the two Intel machines I have (one local, one my 84 year old dad uses) X11 stopped displaying anything after this update. As a quick fix I removed the xorg.conf file completely to get a desktop back. This was a couple of weeks ago. Removing xorg.conf worked on both machines and I haven't gone back to figure out what happened and how to best address it. Good luck, Mark
Re: [gentoo-user] {OT} dedicated server or cloud server?
Cloud services are often far more expensive, I work with someone who did a fair amount of research of the various costs of clouds. They are good for dynamic scaling of resources but if your concentrating on one server or another its likely your server load isn't highly intensive and a single dedicated server could handle it. Also, there are the options of cheaper webhosting, or a VPS, as a true dedicated server can be quite expensive due to the cost of rackspace. In terms of availability, it simply depends on replication and the reliability of the data site. with a standard cloud server there is likely not replication across sites and so the availability is determined by availability of the data center. Dedicated servers dont have multi site replication (unless you do it yourself), however many provide far better uptime SLAs than a cloud provider. For example, Amazon EC2 SLA guaruntees 99.95% uptime. whereas dedicated servers or VPSs can generally offer between 99,99% and 99.% (depending on who it is). -Kevin Brandstatter On Fri, Dec 14, 2012 at 8:34 AM, Florian Philipp li...@binarywings.netwrote: Am 14.12.2012 11:00, schrieb Grant: Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant Last time I did the calculation, a dedicated or normal virtualized infrastructure was more cost effective as long as you could accurately predict the performance you need. Cloud services only really help if you need a high dynamic range regarding scale and performance, e.g. a service that could get a lot of new users very fast or is only really active for short time spans. Doesn't a good cloud server also have potentially higher availability compared to dedicated? - Grant I'd be grateful if anyone can point me at a well conducted study on that topic. Until then I just say that my anecdotal evidence shows the opposite: My cheap-ass virtual server has an uptime of 492 days with only minor, previously announced network outages. During the same time, Amazon EC2 had what, 3 or 4 major outages? Regards, Florian Philipp
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 11:18:21AM -0500, fe...@crowfix.com wrote: I will try some more desperate tricks today, like reconnecting the USB pile to see if it at least boots the disks again - is my choice between disks and keyboard? I will find out. My best guess right now is that booting 3.7.0 is what clobbered things; whether I added a option which loaded bad firmware, or 3.7.0 is broken, I have no idea. It could well be something unrelated to 3.7.0. My goal for today is to try to get keyboard and disk working, then boot with 3.6.8. Whatever you think of logic, it is entirely illogical that a kernel could kill your BIOS, or any hardware ... at least, just booting into it. The southbridge is a good thing to look at, esp for a burned spot/pit. My suggestion is http://www.sysresccd.org/SystemRescueCd_Homepage Our Gentoo images have been broken lately, and are short on tools always. -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 12:16:46PM -0600, Bruce Hill wrote: On Fri, Dec 14, 2012 at 11:18:21AM -0500, fe...@crowfix.com wrote: I will try some more desperate tricks today, like reconnecting the USB pile to see if it at least boots the disks again - is my choice between disks and keyboard? I will find out. My best guess right now is that booting 3.7.0 is what clobbered things; whether I added a option which loaded bad firmware, or 3.7.0 is broken, I have no idea. It could well be something unrelated to 3.7.0. My goal for today is to try to get keyboard and disk working, then boot with 3.6.8. Whatever you think of logic, it is entirely illogical that a kernel could kill your BIOS, or any hardware ... at least, just booting into it. The southbridge is a good thing to look at, esp for a burned spot/pit. My suggestion is http://www.sysresccd.org/SystemRescueCd_Homepage That's what I've been using. But the hardware failure is illogical too; why would USB and SATA fail at the same time? Or why would southbridge fail when it had been running perfectly fine? I don't really think it was 3.7.0, but who knows, did I answer some config question incorrectly and tell it to load some firmware? Without access to the disk, I can't tell. I don't remember any question about loading BIOS firmware, and can't see why the kernel would even care about that. The whole mess makes no sense. -- Felix Finch, a la mode
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 01:24:10PM -0500, fe...@crowfix.com wrote: That's what I've been using. But the hardware failure is illogical too; why would USB and SATA fail at the same time? Or why would southbridge fail when it had been running perfectly fine? I don't really think it was 3.7.0, but who knows, did I answer some config question incorrectly and tell it to load some firmware? Without access to the disk, I can't tell. I don't remember any question about loading BIOS firmware, and can't see why the kernel would even care about that. The whole mess makes no sense. Boot with SystemRescueCd and you can't get to a prompt? If you do, do you have internet access once you're there? If neither, take a photo and post a link to a *clear* snapshot. Yes, your southbridge chipset could just happened to have failed at the same time; or it failed on the reboot; or USB and SATA are both on the southbridge that failed so you lost both, basically. -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] ~amd64 compatibility with modern cpus
Am Freitag, 14. Dezember 2012, 12:16:46 schrieb Bruce Hill: Whatever you think of logic, it is entirely illogical that a kernel could kill your BIOS, or any hardware ... at least, just booting into it. emm, not, it isn't. -- #163933
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 12:34:49PM -0600, Bruce Hill wrote: Boot with SystemRescueCd and you can't get to a prompt? Currently can't even boot -- it hangs wit a blank screen at the point grub or the rescue DVD would take over. Yes, your southbridge chipset could just happened to have failed at the same time; or it failed on the reboot; or USB and SATA are both on the southbridge that failed so you lost both, basically. Then my natural naive question is, can this be readily replaced, or is it soldered in and/or obsolete? It is about 8 years old. -- Felix Finch, a la mode
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 1:43 PM, fe...@crowfix.com wrote: On Fri, Dec 14, 2012 at 12:34:49PM -0600, Bruce Hill wrote: Boot with SystemRescueCd and you can't get to a prompt? Currently can't even boot -- it hangs wit a blank screen at the point grub or the rescue DVD would take over. Yes, your southbridge chipset could just happened to have failed at the same time; or it failed on the reboot; or USB and SATA are both on the southbridge that failed so you lost both, basically. Then my natural naive question is, can this be readily replaced, or is it soldered in and/or obsolete? It is about 8 years old. Almost certainly no. Not unless you take the part numbers in question and hit ebay. -- :wq
Re: [gentoo-user] ~amd64 compatibility with modern cpus
On Fri, Dec 14, 2012 at 01:43:01PM -0500, fe...@crowfix.com wrote: On Fri, Dec 14, 2012 at 12:34:49PM -0600, Bruce Hill wrote: Boot with SystemRescueCd and you can't get to a prompt? Currently can't even boot -- it hangs wit a blank screen at the point grub or the rescue DVD would take over. Yes, your southbridge chipset could just happened to have failed at the same time; or it failed on the reboot; or USB and SATA are both on the southbridge that failed so you lost both, basically. Then my natural naive question is, can this be readily replaced, or is it soldered in and/or obsolete? It is about 8 years old. An 8-year old consumer class mobo begs replacing. I think the way to replace the southbridge chip (if that's it) is called floating point solder, or some such. Hardware fails, and just a new kernel for your present drive will get you up and running on a new board. -- Happy Penguin Computers ') 126 Fenco Drive ( \ Tupelo, MS 38801 ^^ supp...@happypenguincomputers.com 662-269-2706 662-205-6424 http://happypenguincomputers.com/ Don't top-post: http://en.wikipedia.org/wiki/Top_post#Top-posting
Re: [gentoo-user] Re: Anyone switched to eudev yet?
Sorry, you're right, I'll go back to sleep now... ;) I spoke without looking, and indeed my mask is set to =181 On 2012-12-14 12:34 PM, Bruce Hill da...@happypenguincomputers.com wrote: On Fri, Dec 14, 2012 at 11:20:05AM -0500, Tanstaafl wrote: On 2012-12-14 10:39 AM, Bruce Hillda...@happypenguincomputers.com wrote: What Mark wrote you is golden. I might only add that if you put: =sys-fs/udev-181 into /etc/portage/package.mask you will have the present stable udev from*before* those weirdos starting messing it up, forcing systemd to take over udev. Hmmm... For some reason I have masked my udev at 171... Are you saying I can change the mask to 181 and not have to worry about having an unbootable system due to my /usr being on a separate partition? The =sys-fs/udev-181 in /etc/portage/package.mask means that any udev version equal to or greater than udev-181 is masked from the system.
Re: [gentoo-user] ~amd64 compatibility with modern cpus -- REBOOT OK THIS A.M.
I finally ran out of excuses to not reboot after a night powered off, and it did. It's all running normally now, but I think it's time for me to take the hint, grab a clue, and start researching a replacement. -- Felix Finch, a la mode
Re: [gentoo-user] ~amd64 compatibility with modern cpus -- REBOOT OK THIS A.M.
On Fri, Dec 14, 2012 at 2:51 PM, fe...@crowfix.com wrote: I finally ran out of excuses to not reboot after a night powered off, and it did. It's all running normally now, but I think it's time for me to take the hint, grab a clue, and start researching a replacement. I think you'll find the power consumption of modern systems absolutely delightful compared to 8yo systems. I was amazed to see that server-grade xeon CPUs can be had for ~250USD, with a 77W TDP. 5-6 years ago, my Phenom 9650 was a 120TDP part for almost that much when not on sale. (Yes, I know Intel and AMD don't calculate their thermal data quite the same way, but it seems comperable enough.) -- :wq
Re: [gentoo-user] Question about new USE flags for xf86-video-intel
Am 14.12.2012 18:56, schrieb Mark Knecht: On Fri, Dec 14, 2012 at 9:44 AM, Walter Dnes waltd...@waltdnes.org wrote: I'm updating a system that's probably gone 2 or 3 months since its last update. !!! The ebuild selected to satisfy x11-drivers/xf86-video-intel has unmet requirements. - x11-drivers/xf86-video-intel-2.20.13::gentoo USE=dri -glamor -sna -udev -uxa -xvmc The following REQUIRED_USE flag constraints are unsatisfied: any-of ( glamor sna uxa ) This is the first time I've run into these options. A bit of Google detective work turned up some articles... http://www.phoronix.com/scan.php?page=articleitem=intel_glamor_firstnum=1 http://ickle.wordpress.com/2012/06/05/an-evening-with-glamor/ ...which give the impression that SNA is the fastest option overall. The phoronix article also mentioned some instability issues with glamor. I'll probably be going with SNA. Does anybody else have any experience and/or opinions? -- Walter Dnes waltd...@waltdnes.org I don't run desktop environments; I run useful applications I have no idea about performance but I've got udev and sna enabled at this time. +1 gandalf ~ # emerge -pv xf86-video-intel These are the packages that would be merged, in order: Calculating dependencies... done! [ebuild R] x11-drivers/xf86-video-intel-2.20.13 USE=dri sna udev -glamor -uxa -xvmc 0 kB Total: 1 package (1 reinstall), Size of downloads: 0 kB gandalf ~ # Be forewarned, or please report back if you can, but in my case on the two Intel machines I have (one local, one my 84 year old dad uses) X11 stopped displaying anything after this update. As a quick fix I removed the xorg.conf file completely to get a desktop back. This was a couple of weeks ago. Removing xorg.conf worked on both machines and I haven't gone back to figure out what happened and how to best address it. If you don't miss features I say good riddance to xorg.conf. Regards, Florian Philipp signature.asc Description: OpenPGP digital signature
Re: [gentoo-user] Question about new USE flags for xf86-video-intel
On Fri, Dec 14, 2012 at 12:35 PM, Florian Philipp li...@binarywings.net wrote: If you don't miss features I say good riddance to xorg.conf. Regards, Florian Philipp Yeah, I agree in general, but in this case how does one determine that on a remote machine? I run KDE, my dad runs Gnome. How would I know if things are OK, other than making the change and waiting for complaints I guess. Cheers, Mark
Re: [gentoo-user] Re: Anyone switched to eudev yet?
Bruce Hill wrote: On Fri, Dec 14, 2012 at 10:52:12AM -0600, Dale wrote: I have the flu, nasty one at that, and I really don't need to add hal to my list right now. That said, the Doctor called and the blood tests said I was healthy as a horse, other than being sick as a dog. :/ Sort of like software, supposed to work great out of the box but usually doesn't. Imagine me on windoze. O_O Dale :-) :-) P. S. I'm not sure about the living part yet. It's a theory that is yet to be proven. Maybe I can give you a run for your money. I hit most corner cases head on. See this bug https://bugs.gentoo.org/show_bug.cgi?id=447184 Today we have major things going on, for which I need my printer to print. And yesterday it stopped printing. It will crash FF when Print is selected, and won't print from Inkscape, LibreOffice, or Evince. There were already problems but now it just rolled over and died. And, no, it's not the printer ... workstation CAN print with this command: lp -d HP_Officejet_Pro_8500_A910 -o scaling=75 Happy-Penguin-Gymnastics/offers/Drop-and-Shop-text and baruch can print to the printer (192.168.100.10:9100) Hill's Law: it works best when you don't need it I just depend on Murphy's law. It always works for me. lol Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] ~amd64 compatibility with modern cpus -- REBOOT OK THIS A.M.
Michael Mol wrote: On Fri, Dec 14, 2012 at 2:51 PM, fe...@crowfix.com wrote: I finally ran out of excuses to not reboot after a night powered off, and it did. It's all running normally now, but I think it's time for me to take the hint, grab a clue, and start researching a replacement. I think you'll find the power consumption of modern systems absolutely delightful compared to 8yo systems. I was amazed to see that server-grade xeon CPUs can be had for ~250USD, with a 77W TDP. 5-6 years ago, my Phenom 9650 was a 120TDP part for almost that much when not on sale. (Yes, I know Intel and AMD don't calculate their thermal data quite the same way, but it seems comperable enough.) -- :wq I agree. My old rig, AMD 2500+ CPU with about 2Gbs of ram, a couple hard drives and a FX-5200 video card pulled about 400 watts or so, not counting the monitor. My new rig, AMD 4 core 3.2Ghz CPU, 16Gbs of ram, 4 hard drives and a GT-220 1Gb video card and it pulls less than 150 watts and just barely over 200 watts when compiling something and using the drives a lot while doing it. That includes a LCD monitor too. My old rig had a CRT monitor. Power hog big time which is why I didn't add it to the power being pulled. I did some math, my new rig is almost 8 times faster/powerful than my old rig but pulls much less than half the power even when fully loaded. I might add, I think the old rig was idle when I measured that. I bet you could find some really old and cheap mobos, say a year or two old but still new out of the box, and build a cheap system and have a lot more powerful system that pulls less power from the wall. I saw a Gigabyte mobo the other day on newegg for like $40.00. I'm sure the model is a year or two old but still, its better than a 8 year old. Also, memory is cheap for those too. Most likely the CPU would be too. Sure is a weird problem tho. Sounds like something I would run into myself. ;-) Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!
Re: [gentoo-user] Anyone switched to eudev yet?
On Fri, 14 Dec 2012 08:53:35 -0800 Mark Knecht markkne...@gmail.com wrote: I guess the other question that's lurking here for me is why do you have /usr on a separate partition? What's the usage model that drives a person to do that? The most I've ever done is move /usr/portage and /usr/src to other places. My /usr never has all that much in it beyond those two directories, along with maybe /usr/share. Would it not be easier for you in the long run to move /usr back to / and not have to deal with this question at all? It should be moving in the other direction for stability reasons and busybox is no full answer. On OpenBSD which has the benefit of userland being part of it. All the critical single user binaries are in root and built statically as much as possible, maximising system reliability no matter the custom requirements or packages. The way I have it on OpenBSD / ro 100 megabytes and I never need to fsck and can reliably fix all but the most likely problem and snapshot quickly, though there is no need as the kernel is rock solid. /usr ro,nodev ~600 megabytes that I almost never need to fsck even when I pull the plug /usr/local ro,nodev,nosuid All installed packages go here and I can give users the ability to only mount writeable this location. There are other plusses I won't bother going into. All the BSDs and debian stable (old and initramfs) still get's this right with debian suggesting a seperate /usr during install in compliance with the filesystem hiearchical standard and the upcoming draft/version 3, which states the real technical and uptime benefits of a seperate /usr. https://wiki.linuxfoundation.org/en/FHS Unfortunately stability and security often only get's noticed and chosen over other function when it's completely obliterated and has stopped functioning alltogether. When hard worked (including rusty russel) documents like this get ignored when freedesktop.org is given so much credence even though freedesktop.org is actually simply stating opinion without having debate/comments on it's site and in contrast a combined root/usr has no technical benefit not addressed elsewhere (grub etc..) and when the issues in userland are far from insurmountable it is quite worrying and I am grateful to those who have stood up against this and the trend of added complexity into pid1/systemd and early boot. What is also worrying is the recent trends of the kits, udisks dropping features for months to get multiseat and dbus getting everywhere like Windows and RPC. I can take spread out documentation compared to OpenBSD but some of these issues are quite rediculous, I just wish OpenBSD had more devs for KMS and stable updates as it is perhaps due to being a smaller project involving both core userland and kernel and with hard fast goals, far better managed.
Re: [gentoo-user] ~amd64 compatibility with modern cpus -- REBOOT OK THIS A.M.
On Fri, Dec 14, 2012 at 03:19:53PM -0600, Dale wrote: I did some math, my new rig is almost 8 times faster/powerful than my old rig but pulls much less than half the power even when fully loaded. I might add, I think the old rig was idle when I measured that. My next box will be a commodity box. This one is a big tower because I wanted compute power for pictures I was generating, lots of them, and I wanted lots of storage room. Now with 4TB drives so cheap and USB 3 for speed, the storage can all be done by external drives, and I don't need the compute power any more. I do like building computers, in a way, but I don't like the idea of being rushed into building a replacement, and I don't need any custom features. A friend suggested a big lightning storm a month ago might have temporarily scrambled something. It tripped breakers in the house and blew out a battery in the standby generator shed, but I've rebooted twice since that storm without any problems. I guess it will remain a mystery. -- ... _._. ._ ._. . _._. ._. ___ .__ ._. . .__. ._ .. ._. Felix Finch: scarecrow repairman rocket surgeon / fe...@crowfix.com GPG = E987 4493 C860 246C 3B1E 6477 7838 76E9 182E 8151 ITAR license #4933 I've found a solution to Fermat's Last Theorem but I see I've run out of room o
Re: [gentoo-user] {OT} dedicated server or cloud server?
Would everyone here be in favor of a dedicated server over a cloud server from a host with good cloud infrastructure? The cloud server concept is amazing but from what I'm reading a dedicated server at the same price point far outperforms it. - Grant Last time I did the calculation, a dedicated or normal virtualized infrastructure was more cost effective as long as you could accurately predict the performance you need. Cloud services only really help if you need a high dynamic range regarding scale and performance, e.g. a service that could get a lot of new users very fast or is only really active for short time spans. Doesn't a good cloud server also have potentially higher availability compared to dedicated? Potentially? Yes. In reality? No. It's not the virtualization that breaks, it's all the surrounding infrastructure, especially Layer 2. You will not believe how fragile that stuff can get. In the old days, a small slip up could isolate a small part of the network. These days, a small slip-up easily ripples though the entire network and takes down all of it, and sadly this is not rare. The networking needs of VMs are radically different from the traditional, and this is the side-effect: fragility. Sounds like the technology isn't ready to compare favorably with dedicated yet in an ordinary scenario with a website to run. Maybe in a few years? The concept is amazing. I'd also like to move my desktops and even laptops to the cloud once things get solidified. Then client hardware becomes interchangeable, disposable... each physical location would only need one thin client and a bunch of USB peripherals (DisplayLink, etc). - Grant
Re: [gentoo-user] Dual or Quad CPU complications?
So if I have 2 physical CPU's with 4 cores each and I enable SMP, I'm using 8 cores? Can NUMA be either enabled or disabled when using more than one physical CPU, or is it required? NUMA is a hardware architecture. It's how you access memory on a hardware level: NUMA = Non Uniform Memory Access vs a UMA architecture of typical (old/legacy) SMP systems (UMA = Uniform Memory Access). In a UMA system, all the memory belongs to all the sockets. In a NUMA system, each socket has it's own local memory. In modern (x86-64) processors, each socket has it's own memory controller so each socket controls its own local memory. If one socket runs out of memory it can ask another socket to lend him some memory. In a UMA system, no socket has to ask since memory is global and belongs to all sockets so if one socket uses up all the memory ... the rest starve. In NUMA, there's more control over who uses what (be it cores or RAM). If you have a modern dual or quad (or higher #) socket system ... you've got NUMA architecture and you can't get rid of it, it's hardware, not software. So I must enable CONFIG_NUMA for more than one physical CPU, and disable it for only one physical CPU? Yup. But ... Why would you want to disable a socket (CPU)? If you disable a socket (CPU) ... you lose the memory attached to that socket (CPU) not to mention you lose those cores ;) Sure but it sounds like if my system only has one CPU socket, CONFIG_NUMA should be disabled. - Grant A better solution would be to use cgroups or numactl tools to pin a certain process to a set of cores and a memory region. If you really want to deactivate cores (but not the whole socket), you can type: echo 0 /sys/devices/system/cpu/cpu1/online This would deactivate core #1. You can deactivate as many cores as you wish, except for core #0. This can be done without rebooting your server (aka during run time). Your memory will not be affected, but you will have less cores (and theoretically more memory bandwidth). I say theoretically because you always have to benchmark these things with YOUR application (remember logic NEVER applies to real life ;) If you want to check the # of cores you've got: cat /proc/interrupts | grep CPU Other possibilities such as cat /proc/cpuinfo or dmesg, ... can be useful too for this: your choice, FLOSS gives you options. If you want to activate the previously deactivated core, you can run: echo 1 /sys/devices/system/cpu/cpu1/online Now ... be sure your core numbering is the expected core numbering. IOW, not all server vendors follow the same numbering scheme so core #1 in vendor A's server could be core #2 in vendor B's server. Never trust logic ;) As I mentioned previously: test/benchmark YOUR software. DON'T trust logic or generic benchmarks or web pages with results. Trust YOUR results only. HTH Rafa
Re: [gentoo-user] Intel Atom: architecture, distcc, crossdev and compile flags
On Wed, Dec 12, 2012 at 09:16:58AM +0100, Florian Philipp wrote: * The last thing I’m going to set up is filesystem encryption, at least for ~. I already know/think that AES would be the best choice due to limited CPU power, but what else is there to heed besides key size? Nothing, you're good. Hash and key chaining method have negligible impact. If you stick with an x86_32 userspace I suggest at least using an x64 kernel so you can use of CRYPTO_AES_X86_64. That's an interesting idea. [...] I haven't done any comparisons of 32/64 crypto yet, I'm just reading docs on Luks (never used it before). Well now, I did a few comparisons yesterday. Not much---just permutated a few of the most probable crypto combinations (aes/twofish, cbc/xts, essiv/plain). I created the LUKS container, opened it and gave it a spin with hdparm -t. The result was shocking and outrageous; reported throughput w/o encryption was 75 MB/s, which is your typical 5400 rev laptop HDD. First I was disappointed when I saw what aes-cbc-essiv gave me on 32 bit: a mere 19 and a bit. But on 64 bit, it yielded a whopping 34 MB/s. I had a hunch and booted the 32 bit system with the 64 bit kernel---and throughput stayed high as expected. So for the sake of simplicity (and to give it a rest after two weeks of ricing to the day), I will use the 32 bit userland with a 64 bit kernel. I will only need to set up some magic (a multilib crossdev gcc and separate build dirs) so I can build both kernels with their separate configs from the same source dir. I personally see no reason for encrypting root as there is nothing of interest in there. Hm ideed, the only password I have in a plaintext config file are WiFi passwords vor wpa and vpnc. For those the symlink solution could be used. Not needing an initrd is a big incentive. :) On a sidenote, While I was cleaning up unread mails in the ML, I just found your interesting frontswap/zcache trick. I tried that, too, but for now will keep it disabled---simple copying of big files was slowed down to 33 MB/s, obviously b/c the cache is constantly being changed. It's just not suitable for little Atoms. -- Gruß | Greetings | Qapla’ Please do not share anything from, with or about me with any Facebook service. The duration of a minute is relative. It depends on the side of the toilet door you are standing on. signature.asc Description: Digital signature
Re: [gentoo-user] Dual or Quad CPU complications?
On Thursday 13 Dec 2012 14:13:56 Bruce Hill wrote: On Thu, Dec 13, 2012 at 08:44:45AM +0100, J. Roeleveld wrote: NUMA is also an option in the kernel. Should also be fully transparent. I got one machine with NUMA and only had to set an option for it. Does anyone know how to check it's working properly? dmesg | grep NUMA Hmm ... it seems that it can't find NUMA configuration: $ dmesg | grep UMA No NUMA configuration found Am I supposed to configure something in userspace? This is what the kernel has: $ uname -a Linux dell_xps 3.5.7-gentoo #2 SMP PREEMPT Mon Nov 26 10:36:47 GMT 2012 x86_64 Intel(R) Core(TM) i7 CPU Q 720 @ 1.60GHz GenuineIntel GNU/Linux $ cat /usr/src/linux/.config | grep -i NUMA CONFIG_NUMA=y # CONFIG_AMD_NUMA is not set CONFIG_X86_64_ACPI_NUMA=y # CONFIG_NUMA_EMU is not set CONFIG_USE_PERCPU_NUMA_NODE_ID=y CONFIG_ACPI_NUMA=y -- Regards, Mick signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] crontab questions
Thanks Michael. I'd like to have more control over when the commands are run. Maybe the system crontab (cronbase) should be used when that control isn't necessary or to allow programs to add stuff to a crontab, and a user crontab should be used when more control is necessary? I personally like the idea of the cron.{daily,weekly,...}, but the implementation is a little goofy. On our mail server, I've added an additional directory called cron.bihourly to update virus/spam signatures every two hours. The simplest way to accomplish this is to add, # Run every two hours 0 */2 * * * root find -L /etc/cron.bihourly -type f -executable \ -execdir '{}' \; in the global /etc/crontab. I'm sure this is horribly deficient according to whoever implemented the run-crons stuff, but for me the additional clarity is worth it. You can of course add anything else you like in the global/user crontabs, and they'll work normally. OK, I've moved all of my user crontabs (including root) to /etc/crontab. But be careful: do you really want `emerge -puDN` to run 15 minutes after you start an `eix-sync`? Or do you just want it to run when `eix-sync` is done? If it's the latter, you don't want to schedule it 15 minutes later -- you could hit a slow mirror and still be updating when the `emerge` kicks off. In that case it's better to put all of the commands in one script, and schedule that when you want. That way the commands occur in sequence, and you can bail out if something fails. Done. I think it's better for me to pipe the commands to mailx. I get mail if I run this on the command line emerge -pvDuN world | /usr/bin/mail -s subject -a From: from m...@email.com mailto:m...@email.com But I don't get any mail when it runs in the crontab. Do you know why that's happening? I do get mail from 'emerge -pvDuN world' run in the crontab without piping it to mail. I got it working in /etc/crontab. Should I file a bug for http://www.gentoo.org/doc/en/cron-guide.xml to mention that vixie-cron must be restarted when making changes to /etc/crontab? It says: Note that only Vixie-cron schedules jobs in /etc/crontab automatically. I'm not sure. I do the same thing, though, albeit with a temporary file (and it works). Maybe try `echo`ing the output to a file? This script emails me the current iptables to make sure fail2ban hasn't gone berserk: #!/bin/bash # Send the current iptables -L -n output to the postmaster. TMPFILE=/tmp/iptables-state.log MAILADDR=postmas...@example.com echo To: $MAILADDR $TMPFILE echo From: r...@mx1.example.com $TMPFILE echo Subject: mx1 iptables state $TMPFILE iptables -L -n $TMPFILE sendmail -f r...@mx1.example.com \ $MAILADDR \ $TMPFILE rm $TMPFILE It's not very fancy but it does work. If a temp file works for you, it might help you narrow down the problem. Wouldn't you rather use a one-liner like this? iptables -L -n | mail -s mx1 iptables state -a From: r...@mx1.example.com postmas...@example.com - Grant
Re: [gentoo-user] Dual or Quad CPU complications?
So if I have 2 physical CPU's with 4 cores each and I enable SMP, I'm using 8 cores? Can NUMA be either enabled or disabled when using more than one physical CPU, or is it required? NUMA is a hardware architecture. It's how you access memory on a hardware level: NUMA = Non Uniform Memory Access vs a UMA architecture of typical (old/legacy) SMP systems (UMA = Uniform Memory Access). In a UMA system, all the memory belongs to all the sockets. In a NUMA system, each socket has it's own local memory. In modern (x86-64) processors, each socket has it's own memory controller so each socket controls its own local memory. If one socket runs out of memory it can ask another socket to lend him some memory. In a UMA system, no socket has to ask since memory is global and belongs to all sockets so if one socket uses up all the memory ... the rest starve. In NUMA, there's more control over who uses what (be it cores or RAM). If you have a modern dual or quad (or higher #) socket system ... you've got NUMA architecture and you can't get rid of it, it's hardware, not software. So I must enable CONFIG_NUMA for more than one physical CPU, and disable it for only one physical CPU? Yup. But ... Why would you want to disable a socket (CPU)? If you disable a socket (CPU) ... you lose the memory attached to that socket (CPU) not to mention you lose those cores ;) Sure but it sounds like if my system only has one CPU socket, CONFIG_NUMA should be disabled. I read this in make menuconfig: The kernel will try to allocate memory used by a CPU on the local memory controller of the CPU and add some more NUMA awareness to the kernel. For 64-bit this is recommended if the system is Intel Core i7 (or later), AMD Opteron, or EM64T NUMA. To be sure I have this right, I should disable CONFIG_NUMA on any system with a single physical CPU, even if it's an AMD Opteron? - Grant
Re: [gentoo-user] Dual or Quad CPU complications?
On Fri, Dec 14, 2012 at 10:16 PM, Grant emailgr...@gmail.com wrote: So if I have 2 physical CPU's with 4 cores each and I enable SMP, I'm using 8 cores? Can NUMA be either enabled or disabled when using more than one physical CPU, or is it required? NUMA is a hardware architecture. It's how you access memory on a hardware level: NUMA = Non Uniform Memory Access vs a UMA architecture of typical (old/legacy) SMP systems (UMA = Uniform Memory Access). In a UMA system, all the memory belongs to all the sockets. In a NUMA system, each socket has it's own local memory. In modern (x86-64) processors, each socket has it's own memory controller so each socket controls its own local memory. If one socket runs out of memory it can ask another socket to lend him some memory. In a UMA system, no socket has to ask since memory is global and belongs to all sockets so if one socket uses up all the memory ... the rest starve. In NUMA, there's more control over who uses what (be it cores or RAM). If you have a modern dual or quad (or higher #) socket system ... you've got NUMA architecture and you can't get rid of it, it's hardware, not software. So I must enable CONFIG_NUMA for more than one physical CPU, and disable it for only one physical CPU? Yup. But ... Why would you want to disable a socket (CPU)? If you disable a socket (CPU) ... you lose the memory attached to that socket (CPU) not to mention you lose those cores ;) Sure but it sounds like if my system only has one CPU socket, CONFIG_NUMA should be disabled. I read this in make menuconfig: The kernel will try to allocate memory used by a CPU on the local memory controller of the CPU and add some more NUMA awareness to the kernel. For 64-bit this is recommended if the system is Intel Core i7 (or later), AMD Opteron, or EM64T NUMA. To be sure I have this right, I should disable CONFIG_NUMA on any system with a single physical CPU, even if it's an AMD Opteron? No harm done if you enable NUMA on a system where it's not necessary. -- :wq
Re: [gentoo-user] System maintenance procedure?
I think you're right about that. Can I configure eclean to wait a certain number of days since a package was removed before cleaning it? Even if I only run it once per week, it could remove a package that was updated yesterday that I find out I need tomorrow. - Grant -t, --time-limit=timedon't delete files modified since time time is an amount of time: 1y is one year, 2w is two weeks, etc. Units are: y (years), m (months), w (weeks), d (days) and h (hours). I just realized that --time-limit doesn't look like it takes into consideration when a package was removed from the system, only when it was installed. Does anyone know how eclean behaves as far as leaving packages behind for a while in case they're needed? This just got me today. I recently updated google-chrome on one system, 'eclean packages' ran at some point, then chrome started acting up and I couldn't go back to the previous version because eclean had wiped out the package. I don't think we can count on --time-limit to save us because it can still wipe out all previous versions of a package. What we need is a way to keep at least one older version of each package. - Grant
[gentoo-user] /usr/share/doc/openrc/net.example not found
Hello, The file /etc/conf.d/net reports that I can seen an example format at this location: /usr/share/doc/openrc/net.example On my machine that example file does not exist. Did I do something wrong or is this just a documentation oversight? Thank you, Chris PS: I'm trying to find a way to prevent dhcpd from updating my ntp.conf
Re: [gentoo-user] /usr/share/doc/openrc/net.example not found
Chris Stankevitz wrote: Hello, The file /etc/conf.d/net reports that I can seen an example format at this location: /usr/share/doc/openrc/net.example On my machine that example file does not exist. Did I do something wrong or is this just a documentation oversight? Thank you, Chris PS: I'm trying to find a way to prevent dhcpd from updating my ntp.conf Try this: ls /usr/share/doc/openrc-0.11.8/net.example.bz2 You may have to edit the version. Tab completion may be of use on this one. It's how I found it. Hmmm, why is it compressed? Dale :-) :-) -- I am only responsible for what I said ... Not for what you understood or how you interpreted my words!