Improve LAN speed on FC3, home network?

I have ADSL, always on, PPPoE, 3Mbps down and 380Kbps up. I use wshaper to
stop FTP transfers from totally killing my Internet speed. Internet
connectivity is not the issue though. I have 3 computers, FC3 Linux box is
connected via eth0 directly to ADSL modem configured as bridged Ethernet so
that I can use the Roaring Penguin PPPoE for Linux to connect. eth1 goes to
a hub and the hub connects 2 XP computers to each other, shares printer,
and gives Internet connectivity.

My problem is that local network speed is horrible but I only notice this
if I want to transfer a large file to the Linux box. I use samba to make my
XP boxes visible to Linux and make the Linux shares show up in Network
Places. If I try to copy a 700Mb avi or iso file to the linux box, it takes
about 180 minutes to make the copy. I can literally do this faster by
burning CDs and then walking over to the box and copying it to the hard
drive from the CD.

My XP box uses a built in on the motherboard Marvell Yukon Ethernet
Controller and it works fine. I notices that if I try to copy a 700Mb file
to my girlfriend's computer, the speed is the same, but if I try to copy a
700Mb file from her computer to the Linux box, the amount of time it takes
drops in half, about 40 minutes. I updated my Marvell Yukon Ethernet
Controller Driver to version v8.53.1.3 and now have cut the time in half
that it would take to make that transfer, down to about 40 minutes. Still,
I think that is too long a time, something must be wrong that can be fixed.
4-5 minutes I could accept to transfer a 700Mb file but 40 freaking
minutes? That seems to be an inordinate amount of time to me.

Since I now have mine and my girlfriend's computer at the same speed,
perhaps it is the NIC in the Linux box that is not configured right. I have
a Realtek and a 3com card in the box, lsmod shows the drivers loading:

3c59x 44009 0
8139too 29377 0

Not sure which is eth0 and which is eth1. eth1 is the suspected slow
network card. Can someone give me some pointers on how to troubleshoot this
network speed problem and what would be a reasonable amount of time that it
should take to copy a 700Mb avi file across 5 feet of distance from an XP
machine to a Fedora Core 3 machine?

Thank you.
--
~Ohmster
"Read Ohmster" in subject, bypass spam filter.
ohmster /a/t/ newsguy dot com

In comp.os.linux.networking Ohmster <(E-Mail Removed)>:

[ slow network speed on LAN ]

> 3c59x 44009 0
> 8139too 29377 0

> Not sure which is eth0 and which is eth1. eth1 is the suspected slow

See 'lspci' or/and unplug one and see if 'mii-tool' tells you
where the connection is down.

> network card. Can someone give me some pointers on how to troubleshoot this
> network speed problem and what would be a reasonable amount of time that it
> should take to copy a 700Mb avi file across 5 feet of distance from an XP
> machine to a Fedora Core 3 machine?

Presuming 100 Mbit/sec FD on an unloaded LAN 60-120 sec. should
be possible. Depends if you have a quality NIC/Switch/Hub.

Sounds as if you have a duplex mismatch or cable/hub are broken,
check with 'mii-tool' if possible and the output of
'/sbin/ifconfig', these lines are interesting:

UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:58240306 errors:2 dropped:0 overruns:0 frame:1
TX packets:66204598 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3955039934 (3771.8 Mb) TX bytes:1582955248 (1509.6 Mb)

Post them if they don't give you a clue.

--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo (E-Mail Removed) | perl -pe 'y/a-z/n-za-m/'
#bofh excuse 318: Your EMAIL is now being delivered by the USPS.

Michael Heiming <michael+(E-Mail Removed)> wrote in
news:r1vnk3-(E-Mail Removed):

> [ slow network speed on LAN ]
>
>> 3c59x 44009 0
>> 8139too 29377 0
>
>> Not sure which is eth0 and which is eth1. eth1 is the suspected slow
>
> See 'lspci' or/and unplug one and see if 'mii-tool' tells you
> where the connection is down.

[ohmster@ohmster log]$ lspci
00:00.0 Host bridge: VIA Technologies, Inc. VT82C693A/694x [Apollo
PRO133x] (rev c4)
00:01.0 PCI bridge: VIA Technologies, Inc. VT82C598/694x [Apollo
MVP3/Pro133x AGP]
00:04.0 ISA bridge: VIA Technologies, Inc. VT82C596 ISA [Mobile South]
(rev 23)
00:04.1 IDE interface: VIA Technologies, Inc.
VT82C586A/B/VT82C686/A/B/VT823x/A/C PIPC Bus Master ID
E (rev 10)
00:04.2 USB Controller: VIA Technologies, Inc. VT82xxxxx UHCI USB 1.1
Controller (rev 11)
00:04.3 Host bridge: VIA Technologies, Inc. VT82C596 Power Management
(rev 30)
00:0a.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone]
(rev 30)
00:0d.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-
8139/8139C/8139C+ (rev 10)
01:00.0 VGA compatible controller: nVidia Corporation NV34 [GeForce FX
5500] (rev a1)
[ohmster@ohmster log]$

[root@ohmster ~]# mii-tool
eth0: negotiated 100baseTx-FD, link ok
eth1: negotiated 100baseTx-HD, link ok
[root@ohmster ~]#

[root@ohmster ~]# mii-tool
eth0: negotiated 100baseTx-FD, link ok
eth1: no link
[root@ohmster ~]#

eth1 is for sure the link to the hub and the home LAN. Still not sure how
to ID which card it is though so far as brand.

>> network card. Can someone give me some pointers on how to
>> troubleshoot this network speed problem and what would be a
>> reasonable amount of time that it should take to copy a 700Mb avi
>> file across 5 feet of distance from an XP machine to a Fedora Core 3
>> machine?
>
> Presuming 100 Mbit/sec FD on an unloaded LAN 60-120 sec. should
> be possible. Depends if you have a quality NIC/Switch/Hub.
>
> Sounds as if you have a duplex mismatch or cable/hub are broken,
> check with 'mii-tool' if possible and the output of
> '/sbin/ifconfig', these lines are interesting:
>
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:58240306 errors:2 dropped:0 overruns:0 frame:1
> TX packets:66204598 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:1000
> RX bytes:3955039934 (3771.8 Mb) TX bytes:1582955248 (1509.6 Mb)
>
> Post them if they don't give you a clue.


[ohmster@ohmster log]$ ifconfig
eth0 Link encap:Ethernet HWaddr 00:E0:7D:9A:0D:0C
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:172917612 errors:0 dropped:0 overruns:0 frame:0
TX packets:118183742 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:1889579766 (1.7 GiB) TX bytes:4132652440 (3.8 GiB)
Interrupt:9 Base address:0xb800

eth1 Link encap:Ethernet HWaddr 00:50:04:26:A4:62
inet addr:192.168.0.1 Bcast:192.168.0.255 Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:112211987 errors:27 dropped:0 overruns:3 frame:27
TX packets:171190253 errors:0 dropped:0 overruns:0 carrier:0
collisions:700309 txqueuelen:1000
RX bytes:985856594 (940.1 MiB) TX bytes:1827881115 (1.7 GiB)
Interrupt:9 Base address:0xd000

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:16436 Metric:1
RX packets:64524 errors:0 dropped:0 overruns:0 frame:0
TX packets:64524 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:46776217 (44.6 MiB) TX bytes:46776217 (44.6 MiB)

ppp0 Link encap:Point-to-Point Protocol
inet addr:[munge] P-t-P:65.14.252.2 Mask:255.255.255.255
UP POINTOPOINT RUNNING NOARP MULTICAST MTU:1492 Metric:1
RX packets:700397 errors:0 dropped:0 overruns:0 frame:0
TX packets:895511 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:3
RX bytes:135923815 (129.6 MiB) TX bytes:820458261 (782.4 MiB)

Posted. Thank you for your help Micheal. I would like to go further with
this if you have any suggestions.

--
~Ohmster
"Read Ohmster" in subject, bypass spam filter.
ohmster /a/t/ newsguy dot com

Michael Heiming <michael+(E-Mail Removed)> wrote in news:r1vnk3-
(E-Mail Removed):

> Presuming 100 Mbit/sec FD on an unloaded LAN 60-120 sec. should
> be possible. Depends if you have a quality NIC/Switch/Hub.

Hub is a D-Link DSH-5
5 Port 10/100Mbps
Dual Speed Hub

--
~Ohmster
"Read Ohmster" in subject, bypass spam filter.
ohmster /a/t/ newsguy dot com

In comp.os.linux.networking Ohmster <(E-Mail Removed)>:
> Michael Heiming <michael+(E-Mail Removed)> wrote in
> news:r1vnk3-(E-Mail Removed):

>> [ slow network speed on LAN ]
>>
>>> 3c59x 44009 0
>>> 8139too 29377 0
>>
>>> Not sure which is eth0 and which is eth1. eth1 is the suspected slow
>>
>> See 'lspci' or/and unplug one and see if 'mii-tool' tells you
>> where the connection is down.

> [ohmster@ohmster log]$ lspci
> 00:0a.0 Ethernet controller: 3Com Corporation 3c905B 100BaseTX [Cyclone]
> 00:0d.0 Ethernet controller: Realtek Semiconductor Co., Ltd. RTL-> 8139/8139C/8139C+
[..]

> [root@ohmster ~]# mii-tool
> eth0: negotiated 100baseTx-FD, link ok
> eth1: negotiated 100baseTx-HD, link ok
> [root@ohmster ~]#

> [root@ohmster ~]# mii-tool
> eth0: negotiated 100baseTx-FD, link ok
> eth1: no link
> [root@ohmster ~]#

> eth1 is for sure the link to the hub and the home LAN. Still not sure how
> to ID which card it is though so far as brand.

Eth1 should be the Realtek card.

[..]

> [ohmster@ohmster log]$ ifconfig
> eth0 Link encap:Ethernet HWaddr 00:E0:7D:9A:0D:0C
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:172917612 errors:0 dropped:0 overruns:0 frame:0
> TX packets:118183742 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:1000
> RX bytes:1889579766 (1.7 GiB) TX bytes:4132652440 (3.8 GiB)
> Interrupt:9 Base address:0xb800

> eth1 Link encap:Ethernet HWaddr 00:50:04:26:A4:62
> inet addr:192.168.0.1 Bcast:192.168.0.255 Mask:255.255.255.0
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:112211987 errors:27 dropped:0 overruns:3 frame:27
> TX packets:171190253 errors:0 dropped:0 overruns:0 carrier:0
> collisions:700309 txqueuelen:1000
> RX bytes:985856594 (940.1 MiB) TX bytes:1827881115 (1.7 GiB)
> Interrupt:9 Base address:0xd000

Looks good so far, try exchanging the hub with a cross-over cable
and see if this improves things. A usual cable might work though,
some nics auto sense this but there's no guarantee.

Try using ftp, which usually delivers the best performance
without much overhead. Check 'iptraf' while transferring some
large file.

[..]

--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo (E-Mail Removed) | perl -pe 'y/a-z/n-za-m/'
#bofh excuse 270: Someone has messed up the kernel pointers

Michael Heiming <michael+(E-Mail Removed)> wrote in news01ok3-
(E-Mail Removed):

> Looks good so far, try exchanging the hub with a cross-over cable
> and see if this improves things. A usual cable might work though,
> some nics auto sense this but there's no guarantee.

Awww shit man, I just unplugged the hub and then plugged the cable from the
Linux box directly into my XP box, bypassing the hub and the rest of the
network (Forgot about needing a crossover cable.). It worked! I dragged the
same file over to the Linux share and got the same Windows file copy
message with "2 minutes remaining"!

Bad hub I guess. Let me look through my junk and see if I got another one
to try. Hey, now that you mention it, how come it worked when I plugged the
same ethernet cable that went from my Linux NIC directly into my XP NIC. I
am supposed to use a crossover cable for that, aren't I? It did not work
right away, but in like a minute or two, the connection was back up.
Strange. Let me try to swap out the hub and I will report back. Thanks
Michael.

--
~Ohmster
"Read Ohmster" in subject, bypass spam filter.
ohmster /a/t/ newsguy dot com

Michael Heiming <michael+(E-Mail Removed)> wrote in news01ok3-
(E-Mail Removed):

> Looks good so far, try exchanging the hub with a cross-over cable
> and see if this improves things. A usual cable might work though,
> some nics auto sense this but there's no guarantee.

WOW! I went through my box of junk and found a brand new, sealed in the
box, 3com Office Connect 100Base-TX Hub for Networking up to 8 computers.
This thing is nice, way nicer than that crappy little plastic D-Link, has
all kind of lights on it, and man, that MF 700Mb file just flew over there
in about 2 minutes! Gee, I never knew there was anything wrong with the D-
Link hub but this for sure proves it. Now I have a normal network and do
not have to shy away from large transfers anymore.

> Try using ftp, which usually delivers the best performance
> without much overhead. Check 'iptraf' while transferring some
> large file.

Oh, I never knew about iptraf, what a wonderful little utility. I have been
using iftop and that was pretty good but this iftraf is built right in.
Good deal, thanks Mike.

And thanks again for getting me up and running so quickly. You da man!

--
~Ohmster
"Read Ohmster" in subject, bypass spam filter.
ohmster /a/t/ newsguy dot com

In comp.os.linux.networking Ohmster <(E-Mail Removed)>:
> Michael Heiming <michael+(E-Mail Removed)> wrote in news01ok3-
> (E-Mail Removed):

>> Looks good so far, try exchanging the hub with a cross-over cable
>> and see if this improves things. A usual cable might work though,
>> some nics auto sense this but there's no guarantee.

> Awww shit man, I just unplugged the hub and then plugged the cable from the
> Linux box directly into my XP box, bypassing the hub and the rest of the
> network (Forgot about needing a crossover cable.). It worked! I dragged the
> same file over to the Linux share and got the same Windows file copy
> message with "2 minutes remaining"!

Sounds good! RTL based nics aren't that fast, but quite cheap and
usually enough for some home network purposes.

> Bad hub I guess. Let me look through my junk and see if I got another one

That seems the reason, try getting some switch, they are pretty
cheap today.

> to try. Hey, now that you mention it, how come it worked when I plugged the
> same ethernet cable that went from my Linux NIC directly into my XP NIC. I
> am supposed to use a crossover cable for that, aren't I? It did not work
> right away, but in like a minute or two, the connection was back up.
> Strange. Let me try to swap out the hub and I will report back. Thanks
> Michael.

As pointed out, some nics can auto sense this and swap RxD and
TxD connectors on there own.

Glad to hear we could find the problem quickly with help of
enough *data* you provided. ;-)

--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo (E-Mail Removed) | perl -pe 'y/a-z/n-za-m/'
#bofh excuse 205: Quantum dynamics are affecting the transistors

Michael Heiming <michael+(E-Mail Removed)> wrote in
news:jf2ok3-(E-Mail Removed):

Gosh Mike, that was like the fasted problem post, resolve, ever! We
really for sure got the job done.

> Sounds good! RTL based nics aren't that fast, but quite cheap and
> usually enough for some home network purposes.

Realtek? Yeah, they work alright and are very cheap. I think that Office
Depot has them as low as $7 sometimes.

>> Bad hub I guess. Let me look through my junk and see if I got another
>> one
>
> That seems the reason, try getting some switch, they are pretty
> cheap today.

Yeah I hear everyone pushing switches nowadays. What, in layman terms, is
the difference between a hub and a switch Michael? Aren't they the same?
Plug in some computers and they all connect together? For sure this 3com
jobber is a much higher quality product. It feels heavy, looks nice, has
a serious power cord and an inline brick so as not to jam up your AC
outlets. There is also some kind of quality of service lights on it,
looks pretty interesting. Have a look:

http://www.ohmster.com/~ohmster/picture/3comHub.jpg

Man this thing is just dreamy. I have been putting up with this for so
long and never really knew there was a problem other than large transfers
were a "go to bed and get it in the morning" deal. Now the stuff just
flys over there!

>> to try. Hey, now that you mention it, how come it worked when I
>> plugged the same ethernet cable that went from my Linux NIC directly
>> into my XP NIC. I am supposed to use a crossover cable for that,
>> aren't I? It did not work right away, but in like a minute or two,
>> the connection was back up. Strange. Let me try to swap out the hub
>> and I will report back. Thanks Michael.
>
> As pointed out, some nics can auto sense this and swap RxD and
> TxD connectors on there own.

Yeah, I think that is what happened. When I plug the regular hub cable
back in, I get connectivity right away, but when I plugged the straight
ethernet cable directly from computer to computer, it took like 1 minute
but connectivity was back, stronger than ever.

> Glad to hear we could find the problem quickly with help of
> enough *data* you provided. ;-)

Oh for sure. I have been around Usenet long enough to know you really
have to give good information if you want any kind of serious help. I am
so surprised that you just happened to be at your box, ready to help.
This is the second time you did this for me Michael. The first is when
you got our your nvidia box and figured how to update the nvidia drivers
on Fedora for me when there were no kernel headers on the system. Seems
that Fedora stopped releasing the kernel headers and we had to do it with
the src rpm. Many thanks for that.

Well this is for sure a happy story. I updated my Marvell drivers, got
back what is reasonable speed for that built in controller, then with
your suggestions, got suspicious of the hub and there is was. Happy
Memorial Day to you.

--
~Ohmster
"Read Ohmster" in subject, bypass spam filter.
ohmster /a/t/ newsguy dot com

In comp.os.linux.networking Ohmster <(E-Mail Removed)>:
> Michael Heiming <michael+(E-Mail Removed)> wrote in
> news:jf2ok3-(E-Mail Removed):

[ slow LAN transfer resolved, flaky hub ]

> Yeah I hear everyone pushing switches nowadays. What, in layman terms, is
> the difference between a hub and a switch Michael? Aren't they the same?

Not really, this looks like a comprehensive explanation:

"Hub, switch, what is difference?"

http://handsonhowto.com/lan102.html
[..]

> http://www.ohmster.com/~ohmster/picture/3comHub.jpg

Looks nice!

[..]

> Oh for sure. I have been around Usenet long enough to know you really
> have to give good information if you want any kind of serious help. I am

Indeed, sadly this seems to get more and more a problem,
especially with G2 users. Dunno why but some of those feel
insulted pretty fast, just if you ask for required data to
perhaps enable someone to give hints on how to solve some
problem.

The following might be the problem, better written then I could
say it:

http://catb.org/~esr/faqs/smart-questions.html#keepcool

"Much of what looks like rudeness in hacker circles is not
intended to give offense. Rather, it's the product of the direct,
cut-through-the-bullshit communications style that is natural to
people who are more concerned about solving problems than making
others feel warm and fuzzy."

> so surprised that you just happened to be at your box, ready to help.
> This is the second time you did this for me Michael. The first is when
> you got our your nvidia box and figured how to update the nvidia drivers
> on Fedora for me when there were no kernel headers on the system. Seems
> that Fedora stopped releasing the kernel headers and we had to do it with
> the src rpm. Many thanks for that.

In the mean time kmod-nvidia has been added to the livna yum
repositories, did exchange some parts of the box in question (AMD
64) and it runs FC 5 x86_64 now. yum update will install
kmod-nvidia on its own, so there's nothing to do manually
anymore. ;-)

> Well this is for sure a happy story. I updated my Marvell drivers, got
> back what is reasonable speed for that built in controller, then with
> your suggestions, got suspicious of the hub and there is was. Happy
> Memorial Day to you.

Thx.

--
Michael Heiming (X-PGP-Sig > GPG-Key ID: EDD27B94)
mail: echo (E-Mail Removed) | perl -pe 'y/a-z/n-za-m/'
#bofh excuse 165: Backbone Scoliosis