|
NZAmoeba posted:How do I check duplex settings at the client side? (it plugs into a tiny switch). And how do I disable ipv6 lookups? When I was googling this problem I saw a number of people say to disable ipv6 dns lookups in firefox but that wouldn't propagate to the rest of the system would it? (they described a problem where firefox would be slow, but downloads were normal, different from my issue) "ethtool eth0" will tell you if it's full-duplex or not (along with lots of other details). You can disable IPv6 by unchecking the "Enable IPv6 configuration for this interface" checkbox in system-config-network (Devices tab, select the device in question, click Edit). (This isn't causing your problem, though -- it would only make name resolution approximately twice as slow, not actual data transfers.)
|
#
¿
Mar 3, 2009 23:35
|
|
|
|
| # ¿ May 7, 2024 19:17 |
|
|
It is also completely useless unless you're running commands in a terminal. In conclusion, don't read Phoronix.
|
|
#
¿
Nov 18, 2010 22:11
|
|
|
ExcessBLarg! posted:He flipped his poo poo during a license controversy a few years later, and I don't think has been heard of much since. Thankfully, OpenSolaris came along and took one for the team.
|
|
#
¿
Feb 28, 2011 20:39
|
|
|
Misogynist posted:nscd doesn't query anything -- you may be confusing it with SSSD on RHEL6 and recent Fedoras. Here's what NSCD does: This is completely wrong -- nscd is a global cache for all users, so letting any user account populate that cache with results would be a gaping security hole. If nscd is running, glibc asks nscd to do the query, otherwise, glibc does the query directly. (It's actually a bit more complicated than that -- glibc will snoop through nscd's cache, and if it can't find a result in the cache, then it wil ask nscd to the query.) In no situation is anything ever added to nscd's cache by anything other than nscd itself.
|
|
#
¿
Mar 23, 2011 01:26
|
|
|
getaddrinfo() will actually resolve a host:service pair to a prefilled struct sockaddr which can be passed directly to socket() or bind().
|
|
#
¿
Apr 7, 2011 05:51
|
|
|
tee
|
|
#
¿
Apr 8, 2011 09:00
|
|
|
I always use $() because it is easier to see than backquotes.
|
|
#
¿
May 5, 2011 02:16
|
|
|
At least they didn't skip straight to Linux 7 like Sun with both Solaris 7 and Java 5.
|
|
#
¿
May 30, 2011 08:42
|
|
|
ps $(pgrep ntp)
|
|
#
¿
Jun 22, 2011 22:15
|
|
|
systemd integration into the desktop would be limited entirely to the desktop session manager (i.e gnome-session or whatever), which would just make it better on Linux and the same as it was on operating systems which either don't have the equivalent of Linux cgroups or don't have the developer interest to implement it. Which means that Solaris might get something similar, because they both have the kernel support and people paid to care about this stuff, while FreeBSD is still hosed because they can't even get somebody to care about making the X graphics drivers work. pseudorandom name fucked around with this message at 22:18 on Oct 16, 2011 |
|
#
¿
Oct 16, 2011 22:15
|
|
|
Supposedly a lot of the problems that PA had were a direct result of the ALSA drivers being terrible.
|
|
#
¿
Oct 18, 2011 04:01
|
|
|
Wheelchair Stunts posted:This looks to be the command list for version 2. Here is a link to version 4 which requires free registration ( Version 7 is the latest.
|
|
#
¿
Oct 30, 2011 19:17
|
|
) to download.
|
git doesn't store permissions, which is problematic for things like /etc/shadow. etckeeper does, though.
|
|
#
¿
Nov 7, 2011 00:04
|
|
|
Misogynist posted:etckeeper is just a shim around metastore, which is the piece that does all the actual lifting. There are some annoying caveats. It actually doesn't anymore.
|
|
#
¿
Nov 7, 2011 16:04
|
|
|
Phone posted:Did this with my Fedora 16 install (steps 1 & 2)... am I missing something? I know that the anti-aliasing is very subtle but I didn't notice anything change when I installed freetype. Is there a step I'm missing? The TrueType bytecode interpreter patents have expired and Fedora 15 and 16 have that enabled already. The autohinter is ignored if the bytecode interpreter is enabled and the fonts are hinted, which means that step does nothing. freetype-freeworld still exists to enable ClearType subpixel rendering (which is still patented) but you might not notice the effect as much as the bytecode interpreter. And you won't notice any of those nice fonts you've installed unless you've configured them to be used in your desktop or whatever, and those directions don't include that step.
|
|
#
¿
Nov 17, 2011 21:25
|
|
|
Appachai posted:I have a linux work machine, a macbook air, and a windows desktop at home. I'd like to use a portable hard drive to work on things using all these computers. What filesystem do I have to use for all of them to be able to read/write the same drive? Because Microsoft doesn't want anybody else using NTFS. They want you using FAT or exFAT, and they want you paying them licensing fees. FISHMANPET posted:Doesn't Mac support NTFS more natively than that? And NTFS-3G should be in your kernel already, no need to muck around with FUSE. NTFS-3G is a FUSE filesystem, it isn't in your kernel and you need FUSE.
|
|
#
¿
Dec 2, 2011 05:54
|
|
|
The IO schedule chooses which order to send the read or write commands to the disk. The actual location of the sectors read or written is decided by the file system driver (or LVM/device-mapper, the userspace app directly accessing the block device, etc.) Filesystems are designed to minimize fragmentation if possible.
|
|
#
¿
Jan 6, 2012 23:22
|
|
|
The order applications submit requests to the kernel isn't the order they're submitted to the drive. That's the whole point of the IO scheduler, it reorders requests (where possible) to minimize seeks.
|
|
#
¿
Jan 6, 2012 23:42
|
|
|
Shaocaholica posted:Ok, here's a simple case. Copying 10 dirs from one physical drive to another using cp. However, 10 instances of cp are lauched at the same time and allowed to run concurrently. All operations create new files so we don't have to worry about old file locations/sizes. You can expect the IO scheduler to sort the read commands to the source media and the write commands to the destination media (which are in two separate queues) to minimize seeks as much as possible, given the ordering constraints necessary to satisfy the ext3/ext4 integrity guarantees.
|
|
#
¿
Jan 6, 2012 23:56
|
|
|
spankmeister posted:I'm slightly worried about the whole UEFI thing, since it has that "secure boot" thing and all, which effectively means you will need to have a signed bootloader which can only load signed kernels and you cannot sign your own. Yes you can. You can add or remove trusted keys to the UEFI key database, and you can disable secure boot entirely if you so choose. spankmeister posted:Hopefully. You will NOT be able to dual-boot Windows then though. Win8 will rely on the secure boot thing to stop piracy. (Because the most widely used hack for win7 is a bootloader that injects a valid key.) Yes you will. Secure boot explicitly does not require a trusted platform module, so Windows couldn't even tell the difference between a real secure boot and a faked secure boot. But that doesn't matter, because Windows 8 doesn't require secure boot to be enabled in the first place. Secure boot is purely a malware prevention tool. It is a way of making out-of-the-box Windows PCs immune to root kits that insert themselves into the boot process before the OS is loaded.
|
|
#
¿
Jun 7, 2012 09:22
|
|
|
Social Animal posted:Just installed xterm on my server for kicks. I will say it is pretty surreal to run Firefox on my server through SSH. Does anyone use xterm for any serious purpose? It looks like a very ancient and outdated thing, not to mention slow. Social Animal posted:Also if I install a package (let's say emacs) and I want to remove it, doing the yum remove command only removes the package and not the dependencies. What if I want to remove the dependencies as well? Do people ever do that when removing packages or is it better to just keep the dependencies around? yum history undo $TRANSACTION_NUMBER yum install yum-plugin-remove-with-leaves yum remove --remove-leaves $PACKAGE_NAME In either case you'll want to look at the proposed transaction and make sure you're not removing something you need.
|
|
#
¿
Jun 8, 2012 14:31
|
|
|
ToxicFrog posted:Realistically, this sort of thing is often a clusterfuck regardless of the OSes involved; try connecting the same drive to a windows computer and it'll either ignore it completely or offer to format it for you. Because there's no way to make money doing that.
|
|
#
¿
Jul 17, 2012 20:09
|
|
|
Nautilus's SMB support is entirely separate from Samba (the user-space SMB server) or CIFS (the kernel-space SMB filesystem driver).
|
|
#
¿
Jul 17, 2012 22:03
|
|
|
It may, but Windows doesn't use mDNS/DNS-SD for anything, so that wouldn't help.
|
|
#
¿
Jul 17, 2012 22:23
|
|
|
Even if iTunes is installed, Windows Explorer isn't going to use DNS-SD to find network shares, nor is it going to list the local shares using DNS-SD.
|
|
#
¿
Jul 17, 2012 22:34
|
|
|
evol262 posted:Really? That's idiotic. You seriously thought Microsoft would use something not invented in-house?
|
|
#
¿
Jul 17, 2012 23:42
|
|
|
ArcticZombie posted:To clarify I am using Dual Link DVI, I don't know why it's called HDMI2 in xrandr. Those names come from your video card, not your monitor. Are you sure your video card is capable of outputting 2560x1440?
|
|
#
¿
Sep 1, 2012 00:53
|
|
|
No. A single click of the scroll wheel results in a single button 4 or button 5 event, which is up to the application to interpret. And that generally isn't customizable.
|
|
#
¿
Sep 1, 2012 22:16
|
|
|
Oh, lots of people said it was stupid, but they were outnumbered by programs with hard-coded paths.
|
|
#
¿
Sep 5, 2012 21:29
|
|
|
Doctor w-rw-rw- posted:That and like Longinus said, it's implemented as a userspace file system, so the kernel has to make a round trip into and out of userspace rather than kernelspace. I'm not enough of a systems expert to explain why that might hurt performance, but I think that might hurt performance, depending on the workload. Context switches are expensive and the data has to be copied around more than necessary to get it from the original process to disk.
|
|
#
¿
Sep 29, 2012 23:09
|
|
|
Doctor w-rw-rw- posted:Okay, that's what I suspected, but I wasn't confident in it -- What is it about the system and/or its filesystem drivers that make this not an issue? If I recall, context switches cause L1/L2/etc. to have to be cleared in addition to the TLB. Is it just that kernel filesystem drivers live in physical space, giving it a leg up CPU-wise, then? pre:FUSE native
process calling read() process calling read()
| |
v v
kernel FUSE driver kernel filesystem driver
| | data copied
v v
userspace FUSE driver process calling read()
|
v
kernel block driver
| data copied
v
userspace FUSE driver
| data copied
v
kernel FUSE driver
| data copied
v
process calling read()
x86 CPUs are PIPT, so the L1/L2/etc. caches never have to be invalidated on context switch, but the TLB cache does have to be emptied with FUSE, but not with native drivers.
|
|
#
¿
Oct 3, 2012 23:07
|
|
|
Well, feel free to elaborate on why I'm wrong.
|
|
#
¿
Oct 4, 2012 00:10
|
|
|
Netflix realizes there's not enough Linux users to be profitable and there's no way they could make working DRM.
|
|
#
¿
Oct 20, 2012 05:57
|
|
|
The difficulty in getting DRM working on Android is the major reason why Netflix took so long in releasing an Android client.
|
|
#
¿
Oct 20, 2012 06:04
|
|
|
That wouldn't help in the slightest. You can still attach a debugger to a statically linked binary, and you can still modify the kernel to do whatever you want to arbitrary processes.
|
|
#
¿
Oct 20, 2012 07:07
|
|
|
A good-faith DRM effort isn't sufficient, it has to work. DRM was the issue holding Netflix back from Android.
|
|
#
¿
Oct 20, 2012 07:38
|
|
|
ShoulderDaemon posted:Note that this is equally true on the Windows platform. Windows still allows users to install unsigned kernel modules, and custom directvideo drivers, and attach debuggers, and run DMA. Windows is not inherently more DRM-friendly than Linux; Linux just has an arbitrarily higher bar for "secure DRM" than Windows does. DRM capable audio, graphics, and video codec drivers get different WHQL signatures, and can make decisions on what they're willing to do based on what other drivers are loaded. Userspace apps can decide not to talk to non-DRM capable drivers at all.
|
|
#
¿
Oct 20, 2012 23:23
|
|
|
Xenomorph posted:I can't seem to separate the ACLs and POSIX permissions like I can on Mac OS X and FreeBSD. However, Samba+NFS still seems like a workable solution, even with the goofy permissions. I'll just have to check permissions a bit more often on the shares. The "security mask" option doesn't do what you want?
|
|
#
¿
Oct 23, 2012 01:17
|
|
|
Schizophrenic Orb posted:Basically when I start my computer it starts without any graphics, so it puts me at tty1. When I login I have it automatically startX for me. But I don't want tty2-6 to try starting a gui because I already made one through tty1. The normal way to do this is set your distro to boot into graphical mode by default so you don't have to login on a text terminal at all.
|
|
#
¿
Jan 14, 2013 21:57
|
|
|
|
| # ¿ May 7, 2024 19:17 |
|
|
UEFI isn't actually the cause of the problem, just a really easy way to quickly identify which laptops are susceptible to being bricked by the samsung-laptop driver.
|
|
#
¿
Jan 31, 2013 06:24
|
|