| |
| |
| |
|
Page: 1 2 3 4
Comments:
<0> show your tc tunning plz
<1> tc qdisc add dev eth0 root tbf rate 15mbit latency 30ms burst 25kb
<1> that's all.
<1> rather simplistic... but it works on other machines which have the same kernel and almost the verry same debian version (the tc binary is the same at least)
<0> and show your buffers size (per soket)
<0> sysctl -a | grep net.core.
<1> thnx ;-)
<0> sysctl -a | grep net.ipv4.tcp_
<1> I hope pasting all is ok..
<1> net.core.netdev_budget = 300
<1> net.core.somaxconn = 128
<1> net.core.optmem_max = 10240
<1> net.core.message_burst = 10
<1> net.core.message_cost = 5
<1> net.core.netdev_max_backlog = 1000
<1> net.core.dev_weight = 64
<1> net.core.rmem_default = 104448
<1> net.core.wmem_default = 104448
<1> net.core.rmem_max = 104448
<1> net.core.wmem_max = 104448
<1> wheew the tcp stuff is lot's more, hope it's ok to paste that, too
<1> net.ipv4.tcp_abc = 1
<1> net.ipv4.tcp_congestion_control = bic
<1> net.ipv4.tcp_tso_win_divisor = 3
<1> net.ipv4.tcp_moderate_rcvbuf = 1
<1> net.ipv4.tcp_no_metrics_save = 0
<1> net.ipv4.tcp_low_latency = 0
<1> net.ipv4.tcp_frto = 0
<1> net.ipv4.tcp_tw_reuse = 0
<1> net.ipv4.tcp_adv_win_scale = 2
<1> net.ipv4.tcp_app_win = 31
<1> net.ipv4.tcp_rmem = 4096 87380 174760
<1> net.ipv4.tcp_wmem = 4096 16384 131072
<1> net.ipv4.tcp_mem = 49152 65536 98304
<1> net.ipv4.tcp_dsack = 1
<1> net.ipv4.tcp_ecn = 0
<1> net.ipv4.tcp_reordering = 3
<1> net.ipv4.tcp_fack = 1
<1> net.ipv4.tcp_orphan_retries = 0
<1> net.ipv4.tcp_max_syn_backlog = 1024
<1> net.ipv4.tcp_rfc1337 = 0
<1> net.ipv4.tcp_stdurg = 0
<1> net.ipv4.tcp_abort_on_overflow = 0
<1> net.ipv4.tcp_tw_recycle = 0
<1> net.ipv4.tcp_syncookies = 0
<1> net.ipv4.tcp_fin_timeout = 60
<1> net.ipv4.tcp_retries2 = 15
<1> net.ipv4.tcp_retries1 = 3
<0> stop it plz
<1> ok ok
<1> something wrong with that?
<1> I must confess I don't know what most of that means...
<0> wait plz
<1> sure
<2> what I wouldn't mind figuring out is why *my* broadcom network cards periodically freeze the computer. I suspect a bus conflict, but can't actually figure out how to re***ign the IRQ :-/
<3> heyho, could it be that the kernel needs bash for to compile correctly?
<3> seems for me the scripts in ther do not work with ash :/
<4> the kerbnel isc-code not shell-code...
<4> C-code i meant
<3> but the scripts used to build it are shell-scripts i suppose :)
<4> make
<4> oh, for sure
<4> do...
<4> not do, but dough , like homer likes to say ^^
<3> yes, now with bash at least it compiles
<3> took me some hours to find out X]
<4> btw, ash ****s :P
<4> have you seen?
<4> ^^
<3> O_o
<5> you can comile with shells other then bash, you just need too have the proper environment variables & path set which bash dies automagically
<5> & I agree ash ****s it has no history
<4> nothing to say more, ch0rknabe ;)
<3> hm, i use zsh normally
<3> i used ash to replace bash just for scripts
<5> I used to use csh
<4> csh is nice (much c-like)
<4> imho
<5> She sells cshells by the C shore
<2> psh ! :)
<3> but isn't the environment made up by the sripts themselves, except maybe cflags or so?
<4> i just like my little cute bash
<5> ch0rknabe: if an environment is preset & there is nothing too set any part of the environment too something else what happens?
<5> that is basically your issue with ash
<5> it does stuff all, which is what makes it so small & present in most initial ram disks
<3> mh, could you repeat that you said first in other words? :o
<6> is there a reason that vesafb is not in 2.6.15.4?
<6> (or am i just stupid)
<7> hello
<7> I'm progrmaming an app. that rurns in user-space (initramfs), the thing is I might need to print debug messages to dmesg's ring buffer, so I the don't get lost and I can view them after the boot process, can anyone tell me how to proceed?
<7> I've been trying with printk, but it doesn't matter what I do, the appl. doesn't compile, saying "undefined reference to printk"
<8> How can I coax the linux kernel to send system messages (and kernel panics) over a serial cable?
<8> Currently I'm trying this in lilo.conf:
<8> append="console=tty0 console=16550A,io,ttyS0,57600n8"
<9> you're setting the same var twice ?
<10> hi.
<9> hmm i see
<8> from what I'm reading, you can have multiple console= arguements... each one adds a console that the system _should_ print out kernel messages
<9> yep
<8> (to)
<8> using a bash script I found in my search, I can verify that the serial cable works... but the kernel refuses to send kernel messages
<9> does append="console=tty0 console=ttyS0,57600n8" work ?
<8> minicom doesn't report that it is "connected" to another application on boot-up either
<8> nope, I tried that first :)
<9> ah, figured :)
<9> ah, the serial driver is compiled in ?
<8> hmmm, probably not
<10> maybe you could help. Is there a way to add an amount of memory as a write-behind-cache to a single file? (i ask in "kernel" in the hope there could be a special caching-loop-device that could be used.)
<8> theoddbot: of course it had to be something simple. Thanks :)
<8> keck0f_: short of individual file system's cachine or drive caching, I'm not familiar with any
<8> *caching
<9> does anyone know where i should be looking to change the probe order if mount without the -t blah is called ?
<5> you could always use the loop filesystem
<5> probe?
<9> well, how it tries all the different fs
<8> -t or -a?
<9> -t
<9> mount -t vfat vs mount -t ext2
<5> -t will allow you too set it manually
<9> yep, but my kernel isnt sure what its root fs will be
<5> What I did was made two mount points with deifferring fs's for my users
<9> so i cant use rootfstype
<8> I just change my line in the fstab to load floppies (for example) as vfat,msdos
<8> you can set the order there for individual devices
<5> yeah but you can have /mnt/floppy/messydos as msdos, etc
<8> there are limits to that, but I haven't played with it enough to help much more
<9> right, but theres an order, cat /proc/filesystems seems to print that order
<9> and one of my possible root types is that the end
<9> and jffs2 takes ages to decide it cant read the partition
<9> so i want to put it last, somehow
<9> seems the rootfstype=squashfs,cramfs,jffs2 should work
<9> cool :)
<8> live-cd?
<9> game console :)
<8> ah :)
<9> we managed to convince the manufacurer to release the sources after many months
<9> so having a play
<8> nice
<8> 'k, time to test the kernel WITH serial drivers compiled in
<8> thanks again
<8> theoddbot: thanks, compiling the drivers in worked wonders
<9> excellent :)
<9> i've managed to stuff the nand in this thing already :)
<8> 'nand'?
<9> the flash memory in my console
<8> Don't remember hearing that brand name before (yes, I looked it up and realize it is from Micron now)
<9> its just a type of flash
Return to
#kernel
or
Go to some related
logs:
Internal error failed to initialize HAL! ubuntu
inner outer join
gentoo downgrade gcc to 3.4.4
fedora + turn off framebuffer
#php
sid2wav dos
dtrace memleak
compatible java.sql.date with mysql date
daniel drade radetic
gentoo kwlan portage
|
|