Re: [gentoo-user] Re: Best way to improve interactivity with heavy disk activity?
On Sun, Nov 28, 2010 at 7:39 PM, walt wrote: > I'm confused about which of all these various mechanisms apply to single-cpu > machines. AFAICT Con's BFS (e.g.) is really a CPU scheduler and doesn't > affect > single-cpu machines very much. What about CFQ and group scheduling? > Others? > > Thanks for any clues. Don't mix them, CFS --> upstream official CPU scheduler (also supports cgroups, that got used in the 200line patch, which is useless imo) CFQ --> upstream official I/O (disk) scheduler (afaik the only one that supports "ionice") BFS --> ck's CPU scheduler (I don't know what i/o scheduler ck's patchset uses) Anyway, the problem with the long pauses under disk usage is not related to any scheduler at all, and it is a "page cache management" problem that the linux kernel has (in all new versions). CK's patchset includes optimizations to page cache management. (Not related to any scheduler) so it is worth a try. That's it.
Re: [gentoo-user] Re: Best way to improve interactivity with heavy disk activity?
On Sun, Nov 28, 2010 at 2:13 AM, Nikos Chantziaras wrote: > On 11/28/2010 01:03 AM, Stroller wrote: >> >> On 27/11/2010, at 10:22pm, Nikos Chantziaras wrote: >>> >>> The ck patch set does not support group scheduling anyway; >> >> Now I'm a little more confused. Does `ionice` need the CFQ scheduler? > > Nope. I/O scheduling priorities are part of the I/O scheduler, not the CPU > scheduler. CFQ is the official I/O scheduler. ionice only works with CFQ.
Re: [gentoo-user] Best way to improve interactivity with heavy disk activity?
On Fri, Nov 26, 2010 at 5:57 PM, Stroller wrote: > Hi there, > > As per subject, what's the best way to improve interactivity with heavy disk > activity, please? > > Or perhaps a better question would be: what approaches are available? > > Presently my main Linux system is basically just a storage server with a > *really slow* disk controller. I do all my web-browsing and email (and most > other things) on my Mac laptop (because my Mac desktop has recently died ☹), > but I occasionally do some bash or perl scripting, searches and other stuff > on this Linux box. > > Normally this isn't a problem - the machine is an old Pentium 4 but plenty > powerful enough for this simple command-line stuff. However I have recently > bought a new STB which plays DVD .iso files across the network, so I started > ripping DVDs on storage server, using dvdbackup && mkisofs. When I do so, > interactivity becomes *dire* - it takes maybe 15 seconds for *any* command to > execute. > > My immediate reaction was to consider the recent "200-line patch to kernel => > superkernel" thread: > http://www.gossamer-threads.com/lists/gentoo/user/221770 > > But I have also heard of `ionice` in the past: > http://linux.die.net/man/1/ionice > > I've never used that - in fact, I can't recall ever having to use the regular > `nice` - but I think maybe I should consider it. > > Does anyone have any thoughts, please? > > Stroller. > > The "nice" and "ionice" or the cpu "cgroups" are not going to help a lot. The problems of bad interactivity are rarely CPU related so the 200line patch is probably going to do nothing to help. Also "ionice" rarely solves these problems, because the problem is not actually on the i/o scheduler but in the page cache management of the kernel, which reacts pretty bad (awful) under pressure. As far as I know there is nothing you can do to fix interactivity under these loads, except: 1. From getting an SSD, that doesn't actually need page cache :) 2. Fix the page-management/LRU kernel code yourself :) Maybe unofficial kernels (ck,zen) have improved page cache management (I haven't tried them), but the issue still remains unfixed in the upstream kernel.
Re: [gentoo-user] Varying clocks freqs of CPU ?
On Wed, Nov 24, 2010 at 7:00 PM, wrote: > > Hi, > > new features, new problems... > Recent multi-core CPUs modify the clock freqs of their core > if only a subset of all cores are used. > > Are their any CONFIGs need to be set in the linux kernel to > guarantee a stable system running those kind of CPUs ? > > Thank you very much in advance for any help! Most Intel CPUs need: CONFIG_X86_ACPI_CPUFREQ All AMD CPUs need: CONFIG_X86_POWERNOW_K8 enabled in the kernel. And after that you need to install and enable/run the cpufrequtils daemon.
Re: [gentoo-user] Re: State of Radeon drivers
Nvidia's binary can't be compared to ATI's one. The problems you describe are ATI-binary specific. And yes the nvidia binary replaces a lot of Xorg stuff, but after some time you will realise that this is a good thing, as the Xorg is a mess, breaks with updates, and introduces bugs with each release. And because developers know that, they always prepare their software for nvidia, as it is the only *serious* graphics solution for *nix right now. Don't get me wrong, I don't even have an nvidia card in my systems right now (cause ATI are superior in windows, all my systems have ATI), but I miss the times that I had one. So much more stuff worked without problems and with better performance. On Tue, Jul 27, 2010 at 4:42 PM, BRM wrote: > That's great so long as nVidia supports your card. The problem with the > binary drivers is that they typically only support a percentage of all the > cards the video maker makes. > For example, I can't use the ATI binary driver on my laptop since it no > longer supports the R250 chipset, only their latest 3 or 4 generations of > cards. So I have to use the OSS driver, which works great with it. > I have been able to use both the OSS and proprietary drivers on my desktop > with an nVidia card, but I don't know how much longer that will last. > > nVidia's proprietary driver is good namely because it is the same at the > core as on Windows and Mac, and they wrap it to make it work with the *nix > kernels. However, they also do a lot of other funky stuff and keep people > from being able to fully use the full extend of X. Just search this list > (among others) for xRanderer and other components of X and you'll see the > full story of nVidia's proprietary driver. > > Ben > > > *From:* App Deb > *To:* gentoo-user@lists.gentoo.org > *Sent:* Tue, July 27, 2010 5:29:10 AM > *Subject:* Re: [gentoo-user] Re: State of Radeon drivers > > If you are going to use any *nix, nvidia is the best option for years now. > The nvidia closed source drivers are of professional quality and have great > performance. Actually they are the *standard* for graphics in *nix, and many > (professional or not) applications actually support only nvidia. > > The ati oss driver is still under development, sometimes it works ok, > sometimes not, and it is mostly for basic desktop usage and in my opinion it > is progressing too slow. Anyway, I don't like having a driver that uses 10% > of my hardware's capabilties. So until it actually reaches 100% (like the > rest of the linux drivers) I can't recommend ATI on linux and nvidia is the > way to go. > > On Mon, Jul 26, 2010 at 7:32 PM, Florian Philipp < > li...@f_philipp.fastmail.net> wrote: > >> Am 26.07.2010 01:01, schrieb James: >> > Florian Philipp f_philipp.fastmail.net> writes: >> > >> > >> >> I have a quick question: I plan to buy a notebook with an ATI Mobility >> >> Radeon HD 4250. How well would that one work? Can I reasonably expect >> >> Suspend2Ram, 3d acceleration etc to work stable? >> > >> > Well, lots of good information previously posted. Here's a >> > few more tidbits. When ATI video get's older, there's >> > always good opensource solutions to keep using it. Nvidia, >> > sometimes you toss in garbage can, or use vesa or >> > get lucky? Dunno, as I personally avoid Nvidia; other >> > insist on Nvidia. kinda a religious thing with some. >> > >> >> Hehe, religious is the right word. I remember a situation at my >> workplace: The admin of our departement IT ordered a Linux workstation >> with (fully supported) ATI graphics. At the last second he was overruled >> by the head of our institute's IT in favor of a completely unsupported >> and more expensive NVidia card. Not only did the poor guy have to wait >> two more weeks for the shipment to arrive, he was also stuck with the >> VESA driver for half a year and unstable NVidia drivers ever since. >> >> Well, thanks everyone who answered! Problem solved. >> >> Florian Philipp >> >> >
Re: [gentoo-user] Re: State of Radeon drivers
If you are going to use any *nix, nvidia is the best option for years now. The nvidia closed source drivers are of professional quality and have great performance. Actually they are the *standard* for graphics in *nix, and many (professional or not) applications actually support only nvidia. The ati oss driver is still under development, sometimes it works ok, sometimes not, and it is mostly for basic desktop usage and in my opinion it is progressing too slow. Anyway, I don't like having a driver that uses 10% of my hardware's capabilties. So until it actually reaches 100% (like the rest of the linux drivers) I can't recommend ATI on linux and nvidia is the way to go. On Mon, Jul 26, 2010 at 7:32 PM, Florian Philipp < li...@f_philipp.fastmail.net> wrote: > Am 26.07.2010 01:01, schrieb James: > > Florian Philipp f_philipp.fastmail.net> writes: > > > > > >> I have a quick question: I plan to buy a notebook with an ATI Mobility > >> Radeon HD 4250. How well would that one work? Can I reasonably expect > >> Suspend2Ram, 3d acceleration etc to work stable? > > > > Well, lots of good information previously posted. Here's a > > few more tidbits. When ATI video get's older, there's > > always good opensource solutions to keep using it. Nvidia, > > sometimes you toss in garbage can, or use vesa or > > get lucky? Dunno, as I personally avoid Nvidia; other > > insist on Nvidia. kinda a religious thing with some. > > > > Hehe, religious is the right word. I remember a situation at my > workplace: The admin of our departement IT ordered a Linux workstation > with (fully supported) ATI graphics. At the last second he was overruled > by the head of our institute's IT in favor of a completely unsupported > and more expensive NVidia card. Not only did the poor guy have to wait > two more weeks for the shipment to arrive, he was also stuck with the > VESA driver for half a year and unstable NVidia drivers ever since. > > Well, thanks everyone who answered! Problem solved. > > Florian Philipp > >
Re: [gentoo-user] libvirt needs VirtualBox XPCOMC
You should fill a bug then, if you are sure. On Mon, Jul 12, 2010 at 4:01 PM, Crístian Viana wrote: > Do you want to use libvirt with virtualbox? > > > well, I wanted to make some tests with libvirt and VirtualBox, but I guess > I'll leave it for some other time. I removed the "virtualbox" USE flag. > > Else try emerging virtualbox-ose or virtualbox-bin befire libvirt. > > > nothing's changed. > > But this is probably a bug? > > > I believe so. emerge should never fail, in my opinion. if the ebuild lacks > a dependency, it should be pulled; if the code doesn't compile, use patches; > if it has more than one conflicting USE flags, warn the user and use a > default flag; anyway, it should never fail, specially like that. > > :-) > > On Wed, Jul 7, 2010 at 1:11 PM, App Deb wrote: > >> Do you want to use libvirt with virtualbox? If not (say you want to use it >> with kvm or something else) try emerging libvirt without the virtualbox flag >> using package.use . >> >> Else try emerging virtualbox-ose or virtualbox-bin befire libvirt. >> >> But this is probably a bug? libvirt should have virtualbox as a dependency >> if the use flag is enabled, I think. >> >> On Wed, Jul 7, 2010 at 2:32 AM, Crístian Viana >> wrote: >> >>> hi, >>> >>> when I was trying to emerge the newest libvirt ebuild, the following >>> error appeared on configure: >>> >>> checking for VirtualBox XPCOMC location... not found >>> configure: error: VirtualBox XPCOMC is required for the VirtualBox driver >>> >>> what do I need to install to make it work? I'm emerging with the USE >>> flags: >>> >>> [ebuild U ] app-emulation/libvirt-0.8.2 [0.8.1-r1] USE="libvirtd lxc >>> network nls python qemu virtualbox -avahi -caps -iscsi -lvm -macvtap% -nfs >>> -numa -openvz -parted -pcap% -phyp -policykit -sasl (-selinux) -udev -uml >>> -xen" >>> >> >> >
Re: [gentoo-user] Virtualbox USB w/o usbfs
You should never add that line in fstab in Gentoo. Don't follow random tutorials about random distro workarounds. Gentoo bootscripts mount usbfs on boot with gid=group=usb, so if you want to give user permissions for usbfs, you add yourself to the "usb" group, you don't need to do anything else, and Virtualbox will work (of course you also need to be in the vboxusers group). Anyway, usbfs is not actually needed anymore, it has been replaced by usb /dev/ entries. VirtualBox will use usbfs if it is mounted, but if it is not, usb will work with the new improved way (better permission handling, for printers for example). I don't now of any application that works only with usbfs anymore, maybe it is time to remove it (or make it optional) from the default gentoo boot-scripts. On Tue, Jul 13, 2010 at 5:28 PM, SpaceCake wrote: > Hi, > > I have a problem with accessing usb devices from virtualbox after some > kernel/system update. Previously it was ok after I've added the following > line to the fstab > > usbfs /proc/bus/usb usbfs > devgid=1017,devmode=664 0 0 > > but now, there is no /proc/bus/usb > > how can I enable USB access for a regular user in this new situation? > Should I modify some kernel parameters? should I enable CONFIG_USB_DEVICEFS > or there is some alternative solution? > > Thanks > Laszlo > >
Re: [gentoo-user] libvirt needs VirtualBox XPCOMC
Do you want to use libvirt with virtualbox? If not (say you want to use it with kvm or something else) try emerging libvirt without the virtualbox flag using package.use . Else try emerging virtualbox-ose or virtualbox-bin befire libvirt. But this is probably a bug? libvirt should have virtualbox as a dependency if the use flag is enabled, I think. On Wed, Jul 7, 2010 at 2:32 AM, Crístian Viana wrote: > hi, > > when I was trying to emerge the newest libvirt ebuild, the following error > appeared on configure: > > checking for VirtualBox XPCOMC location... not found > configure: error: VirtualBox XPCOMC is required for the VirtualBox driver > > what do I need to install to make it work? I'm emerging with the USE flags: > > [ebuild U ] app-emulation/libvirt-0.8.2 [0.8.1-r1] USE="libvirtd lxc > network nls python qemu virtualbox -avahi -caps -iscsi -lvm -macvtap% -nfs > -numa -openvz -parted -pcap% -phyp -policykit -sasl (-selinux) -udev -uml > -xen" >
Re: [gentoo-user] (gnome) default pdf viewer
This is installed from wine applications, and yes unfortunately they can create a mess (like making the wine "notepad" the default text editor). You need to delete those mime applications that wine creates under your home folder. Try something like: rm -rf $HOME/.local/share/applications/wine or search and delete the file that you told me (wine-extension-pdf). The previous gnome instructions I told you should have worked though. On Wed, Jul 7, 2010 at 4:25 PM, Helmut Jarausch < jarau...@igpm.rwth-aachen.de> wrote: > On 07/07/10 12:45:05, App Deb wrote: > > Right click on a pdf file, select properties -> open with -> and > > select > > evince instead of acroread. > > > > That it the way to easily change default applications for gnome. > > Unfortunately, not for me (Gnome 2.30) > Before and after the above procedure I have > > gnomevfs-info MyFile.pdf > still shows > MIME type : application/pdf > Default app : wine-extension-pdf.desktop > > I wonder where this 'wine-extension-pdf.desktop' is coming from. > > And Balsa tries to use this when viewing a pdf attachment. > I'm lost, > Helmut. > > > > > On Mon, Jul 5, 2010 at 11:53 AM, Helmut Jarausch < > > jarau...@igpm.rwth-aachen.de> wrote: > > > > > Hi, > > > > > > I've discovered balsa , a great mail client, btw. > > > How can I configure the application which is called for a pdf > > > attachment. > > > > > > I cannot see any configuration item for balsa itself, so I suppose > > > it must be a Gnome setting. > > > > > > Currently it's set to acroread, but I'd like to set it evince. > > > I've checked ~/.mime.types and several files in /usr/share/mime* > > > > > > The only file I've found is > > > /usr/share/mime-info/gnome-vfs.keys > > > > > > Is that the right place? How should I edit it (with a standard text > > > editor?) > > > > > > Many thanks for your help, > > > Helmut. > > > > > > > > > > > > > > > >
Re: [gentoo-user] (gnome) default pdf viewer
Right click on a pdf file, select properties -> open with -> and select evince instead of acroread. That it the way to easily change default applications for gnome. On Mon, Jul 5, 2010 at 11:53 AM, Helmut Jarausch < jarau...@igpm.rwth-aachen.de> wrote: > Hi, > > I've discovered balsa , a great mail client, btw. > How can I configure the application which is called for a pdf > attachment. > > I cannot see any configuration item for balsa itself, so I suppose > it must be a Gnome setting. > > Currently it's set to acroread, but I'd like to set it evince. > I've checked ~/.mime.types and several files in /usr/share/mime* > > The only file I've found is > /usr/share/mime-info/gnome-vfs.keys > > Is that the right place? How should I edit it (with a standard text > editor?) > > Many thanks for your help, > Helmut. > > >
Re: [gentoo-user] Why does high-res video drop frames at 60% CPU?
You have dual core so 60% means: 50% (full one core) is for decoding, and the rest 10% is for audio, resizing etc. You can't play the video correctly because your "decoder" is not multithreaded and uses just the one CPU at its fullest. Try using multithreaded version of mplayer "mplayer-mt" (in some overlay probably) with "lavdopts=threads=2" in mplayer config. On Wed, Jul 7, 2010 at 5:17 AM, Grant wrote: > I've been using VDPAU acceleration to play back Blu-Ray rips for a > while, but the extra layer is getting to be quite a hassle so I'm > trying to get decent performance via software decoding. It has > actually come a long way since the last time I tried and playing > Blu-Ray rips via mplayer is nearly watchable. I'm using a dual-core > 3.1Ghz CPU and one of the cores is only taxed up to 60% during > playback, but frames are still being dropped constantly. Does anyone > know where the bottleneck might be? > > - Grant > >
