On Saturday 21 October 2006 13:33, Gleb Smirnoff wrote:
= We aren't currently speaking about performance, we need to know whether
= kernel with DEVICE_POLLING option makes NIC work stable.
Having noticed today's em-driver update, I rebuilt world/kernel and tried the
dump-test again.
The kernel ha
понеділок 23 жовтень 2006 13:37, Mikhail Teterin написав:
> > We aren't currently speaking about performance, we need to know whether
> > kernel with DEVICE_POLLING option makes NIC work stable.
>
> Yes, that seems to be the case...
I spoke too soon :-( It took a lot lo
субота 21 жовтень 2006 13:33, Gleb Smirnoff написав:
> We aren't currently speaking about performance, we need to know whether
> kernel with DEVICE_POLLING option makes NIC work stable.
Yes, that seems to be the case... After I got to the machine's console
(there was no network access) and turning
On Saturday 21 October 2006 11:39, Kris Kennaway wrote:
= We've been discussing em issues for several weeks now, so it would be
= great if you could get yourself up to speed - please review the
= discussion on freebsd-stable and freebsd-net (start with posts by
= Scott Long, myself, and Jack Vogel)
= I'd appreciate if people who are observing the problem will report
= whether adding DEVICE_POLLING option to kernel config helps them
= or not. This will help to tell whether the problem is in the above
= quote or in the import of new versions from vendor.
I tried this yesterday -- before writin
Hello!
The system is a dual Opteron 244 running today's FreeBSD-6.2/amd64.
em-interface connects it to the switch (in gigabit mode).
When I direct 2 database dumps at the machine in parallel (the arriving data
is getting compressed and written to local disk), the "system" component of
the load
четвер 20 липень 2006 14:11, Jack Vogel написав:
> Similar work was done for Linux and is in the 2.6.18 kernel.
>
> We see as much as a 20% improvement in CPU utilization
> when doing a sustained iperf test.
>
> We are hoping to get this code into CURRENT soon if there
> is interest.
There is. I'm
середа 19 липень 2006 22:50, David G. Lawrence написав:
> sendfile() could be extended to allow arbitrary file descriptor types as
> the source and destination, but the zero-copy nature of it can only work
> in the file to socket direction. This is because network buffers can be
> made out of fi
Hello!
My program receives data from the socket and writes it to a file -- with the
usual read()/write() tedium.
Is there anything zero-copying like sendfile() for the socket->file direction?
In fact, sendfile's API may allow to use it in any direction, but the manual
is quite explicit, that t
понеділок 17 липень 2006 15:16, James Ryan написав:
> If you don't mind waiting until the file finishes transferring to
> examine it
But I do, actually. The files we deal with measure in gigabytes, and take a
while to transfer even over fat pipes. Rejecting the uncompressed ones right
away is be
понеділок 17 липень 2006 13:51, David J. Orman написав:
> The stock ftp server?
BTW, what is the stock ftp server on 6-stable? I see two -- ftpd and lukemftpd
and both are installed...
Is there a web-page with comparision somewhere, perhaps? Thanks!
-mi
_
понеділок 17 липень 2006 14:09, David J. Orman написав:
> That could lead to many DoS attacks, high load, etc - but as you said you
> trust the users, I suspect this is not an issue to you. I personally code
> with security in mind no matter the situation, but you decide what is best
> for you. :)
понеділок 17 липень 2006 13:51, David J. Orman написав:
> The stock ftp server? If you can't base the prohibitions on file extension
> alone (such as the 100kb example you made) then you're going to have to
> modify the source of the ftp daemon yourself. Size, extension, etc - those
> are relativel
Hello!
We run an FTP server for the customers to upload their data (usually -- giant
core-files and database-dumps).
Sometimes they forget compress them, however, wasting many gigabytes of our
server's space...
How hard would it be to make the stock FreeBSD FTP-server to examine the
first, sa
понеділок 10 липень 2006 13:50, Brian Candler написав:
> Well, it's probably worth send-pr'ing it.
The rcmdsh() is taken from OpenBSD, I think, and has no room for the stderr.
One would need to reimplement something like rcmdsh2() first :-)
> I'd first test whether rsh itself forwards stderr pro
On Monday 10 July 2006 10:17, Simon L. Nielsen wrote:
= Actually it is, but it would obviously be a stupid idea to do so any
= place where privileged ports are required...
It would be. But where they are NOT required, it is stupid to check the
geteuid() inside the client's rcmd :-)
Thank you ver
The manual page says, that rcmd() is only to be used by root's processes.
On other OSes (Solaris, AIX), trying to call rcmd() without being root simply
fails.
FreeBSD, however, tries to be helpful and invokes rcmdsh in this case, which
is inefficient and leaves the stderr's filedescriptor (fd2p
Why does not the following work?
rem = rcmd(&host, sp->s_port, pwd->pw_name, user, copyargs(argv),
&remerr);
if (rem < 0)
exit(EX_UNAVAILABLE);
if (remerr < 0)
warn("Can't establish stderr channel (%d)", remerr);
It only works, the specif
I'm wondering, if there exists a secure equivalent to rcmd/rexec? Perhaps,
somewhere in libssh?
I need to send data to a command line on another machine, but popen-ing an ssh
session seems like a rather inferior method, because there is no way to
(portably) access the command's stderr...
Thank
After I added the following rules to my ipfw configuration:
ipfw pipe 1 config bw 20Mbit/s
ipfw add 200 pipe 1 tcp from any to any
The following test from Tcl's regression-test suite started to fail:
set s [socket -server accept 0]
set sock ""
set s2 [socket 1
вівторок 27 червень 2006 15:05, Pawel Worach написав:
> > I just noticed, that on my recent "6.1-STABLE #4: Thu Jun 8" amd64
> > system attempts to connect to a bogus port (like ) hang instead of
> > failing with "Connection refused" immediately, as they on other systems.
>
> Using sysctl net.
вівторок 27 червень 2006 15:07, Andrew Pantyukhin написав:
> My first thought is net.inet.tcp.blackhole sysctl.
Yep, set to 2... I wonder, who did this here and why -- the machine is on the
LAN behind firewalls... One of the TCL's http tests is failing because of
this.
Thanks a lot, Andrew and
Hello!
I just noticed, that on my recent "6.1-STABLE #4: Thu Jun 8" amd64 system
attempts to connect to a bogus port (like ) hang instead of failing
with "Connection refused" immediately, as they on other systems.
Why would this be? There is nothing listening: ``netstat -n | grep '' is
e
п'ятниця 09 червень 2006 19:57, Brad написав:
> There is a mistake on that web-page. None of Broadcom's PCI Express
> chipsets support Jumbo frames.
Indeed. Our bge(4) manual page says:
The BCM570x also supports jumbo frames, which can be configured via the
interface MTU setting. Selec
Hello!
I have a bge card, that's identified as:
bge0: mem
0xfe8f-0xfe8f irq 16 at device 0.0 on pci2
According to
http://h18000.www1.hp.com/products/quickspecs/12131_div/12131_div.HTML,
the BCM5751 support jumbo frames up to 9Kb.
However, when I try to increase the MTU on the card ev
On Saturday 01 April 2006 11:19 pm, Luigi Rizzo wrote:
= It would help if you could run, when the traffic stalls, a tcpdump on
= both source and destination, and tell me where you see the traffic.
Ok here is an even simpler case (mspiggy is a Windows 2003 Server):
% ipfw pipe 1 config bw
On Saturday 01 April 2006 11:19 pm, Luigi Rizzo wrote:
= > It is a FreeBSD/amd64-6.1 as of February 7, running on a signle Opteron
= > 244 (hence UP). Machine has 2Gb of RAM and the active interface is em0 in
= > full
=
= so the pipe acts just on traffic from/to the local host ?
No, as I wrote b
On Saturday 01 April 2006 03:57 pm, you wrote:
= i don't know on which version of freebsd is this occurring,
= it would help knowning - as well as knowing if this is an
= UP/SMP and whether it is working as a bridge or router.
It is a FreeBSD/amd64-6.1 as of February 7, running on a signle Opteron
четвер 30 березень 2006 17:06, Luigi Rizzo написав:
> If you are doing it a lot more often, you should probably
> also consider the effect of such frequent changes to the
> pipe's configuration - e.g. pipes respond with a delay
> which is inversely proportional to the bandwidth, so in
> many cases
Hi!
I'm writing an application that, needs to be able to quickly alter the
bandwidth between another machine and the host.
The only way I can do that -- without another machine's cooperation -- is by
using the firewall, such as the dummynet functionality of ipfw.
Is there any way to create/alt
Hello!
I was experiencing serious problems with web-browsing on this one
machine -- various sites would sometimes be very slow or timeout
altogether.
A Windows machine on the other desk is plugged into the same
NAT-router going over to the same cable modem. It is running the
same version of Firef
Hi, Quinton!
Friday, April 22, 2005 01:34, Q wrote:
> Err yes it is. My firewall/router at home has an uptime of 28 days and
> has 0 errrors. Looks like a duplex problem, which will explain your
> poor performance. Why it's happening is another story. Have you tried
> this motherboard using linux
> > I just noticed, that uploading a file over a LANG (at around
> > 5.7Mb/s) resulted in around 25% CPU consumption by the ftpd.
> >
> > I think, that's unusual for a Pentium4 -- what is the process doing?
>
> Check the client does not use ascii mode when uploading (getc() vs
> read()).
That's q
Hi!
I just noticed, that uploading a file over a LANG (at around 5.7Mb/s) resulted
in around 25% CPU consumption by the ftpd.
I think, that's unusual for a Pentium4 -- what is the process doing?
The machine is running 5.2.1-RELEASE and has TrustedBSD extensions.
-mi
___
> On Wed, Apr 20, 2005 at 03:03:57PM -0400, Mikhail Teterin wrote:
> > Well, simply checking out sys/dev/nve, sys/contrib/dev/nve, and
> > sys/modules/nve allowed me to build the if_nve module. Loading if_nve.ko
> > attaches to the interface, but attempts to ifconfig it result
> > The latest version from Q's site works (!), but seems limited to
> > 100baseTX full-duplex. Even at that nominal speed, the actual data
> > transfers seem limited to 2-3 megabytes/s with occasional latency-spikes
> >
> > :-(
> I just tried the one from Q's site on my Asus K8NNXP nForce MCP3
>
> > > > > Try using the tarball from here http://www.onthenet.com.au/~q/nvnet
> > > > Yes! That -- combined with the
> > > > ftp://download1.nvidia.com/XFree86/nforce/1.0-0301/ -- worked!
> > ..
> > > > Could update the net/nvnet port? When 5.4 gets released, I
> > > > imagine plenty of people to b
> > I just installed a fresh 5.4-RC2 on IWill's DK8N motherboard:
> >
> > http://www.iwillusa.com/product_2.asp?p_id=27
> >
> > I transfered the net/nvnet's distfiles using a floppy and built the
> > driver.
> >
> > Unfortunately, kldload-ing if_nv did not make the interface appear...
> > Put
Hi!
I just installed a fresh 5.4-RC2 on IWill's DK8N motherboard:
http://www.iwillusa.com/product_2.asp?p_id=27
I transfered the net/nvnet's distfiles using a floppy and built the driver.
Unfortunately, kldload-ing if_nv did not make the interface appear... Putting
``if_nv_load="YES"''
Hello!
Most of our hosts can only do the regular 1500-byte frames, but some are
Jumbo Frames capable.
I'm trying to make these few servers talk to _each other_ using bigger
frames (the switch supports them) without breaking the LAN into subnets.
In the past someone suggested, I try explicit -mt
Hello!
I'm using the `simple' template in /etc/rc.firewall to allow LAN to access
the Internet from behind the firewall (FreeBSD-stable).
There is a rule there:
# Allow DNS queries out in the world
${fwcmd} add pass udp from any to any 53 keep-state
and, indeed, the firewall mach
[Now CC-ing Chuck Cranor -- the en's author]
=> => http://store.yahoo.com/softbuyweb/inpcidslmod3.html
=> =What makes you think we have a driver for this?
=> The en(4) manual page and the description of this product (on the
=> page above) as one based on Efficient Network's chi
Hello!
I'm trying to make the FreeBSD 5.x machine accept users based on
Kerberos' tickets.
The telnet and telnetd seem like the most functional pair for this
excercise. (rlogin's man page documents Kerberos options, but they
are not implemented, it seems).
The KDC is a Win2K server (with Active
Hello!
The define is ifdef-ed inside __BSD_VISIBLE on FreeBSD. It is not at all
present in RedHat-8's headers. Its description seems to suggest, it is
redundant, as the kernel should be using loopback if possible anyway,
should not it? Can someone clear it out? Thanks!
-mi
To Unsubscrib
Hi!
I'm trying to make it easier to use NFS from my desktop with amd.
Using the vanilla /etc/amd.map (tried changing vers to 2) and only
modifying one of the arguments to amd from the default provided by
/etc/defaults/rc.conf, I have the amd running as:
amd -a /var/run/.amd -l syslog /hos
On Tuesday 04 February 2003 06:44 pm, Wes Peters wrote:
= On Tue, 2003-02-04 at 08:42, Mikhail Teterin wrote:
= > On Monday 03 February 2003 08:19 pm, Wes Peters wrote:
= > = On Tue, 2003-02-04 at 05:27, Mikhail Teterin wrote:
= > = > Hi!
= > = >
= > = > This question
On Monday 03 February 2003 08:19 pm, Wes Peters wrote:
= On Tue, 2003-02-04 at 05:27, Mikhail Teterin wrote:
= > Hi!
= >
= > This question bothered me for a while -- most of the traffic on my
= > LAN is just that -- local. Yet my gw/firewall machine only has one
= > interface
vert socket.
The trick was to figure out, what could be skipped, and what could not.
I'm wondering, if I got that right -- it seems to work find, but does it
leave something open? Before I can recommend it to others, I'd like to
be more sure :-)
-mi
> - Original Message
Hi!
This question bothered me for a while -- most of the traffic on my LAN
is just that -- local. Yet my gw/firewall machine only has one interface
-- with two IP addresses -- private and public on it.
The DSL modem is plugged into the switch just like everything else.
I doubt this is a unique s
Hello!
I set things up once some time ago for one of my machines to relay
e-mail from another -- based on SSL-certificate presented. I'm my
own issuer. The setup was working for a while, but broke recently --
the relay-to-be now rejects relaying, even though it verifies the
certificate Ok.
Here a
Hello!
Rlogin and/or ssh into my machine takes forever. Apparently -- a name
resolving issue, since rlogind remains idle for most of the time. Attaching
debugger to the server rlogind process yields:
0x280ba50f in kevent () from /usr/lib/libc.so.5
(gdb) where
#0 0x280ba50
[Moved to -net]
On 3 Feb, Rodney W. Grimes wrote:
> Almost everyone forgets to configure a reject route for 127/8, ie
> route add -net 127 127.0.0.1 -reject
[...]
> IIRC this route was automatically created at one time in BSD history,
Is not /etc/defaults/rc.conf a good place for this now? L
Hello!
In our load tests we seem to be maxing out the 100Mb full duplex network
card (fxp0).
The machine has two such cards on the motherboard. How can we use both
of them transparently?
I connected the other one (fxp1) to the switch and enabled bridging:
sysctl -w net.link.ether.bri
53 matches
Mail list logo
