Slow ftp send but fast get

I am running a wu-ftp server on redhat 6.2 on a 90mhz pentium with 72mb
of ram. Recently sends to the FTP server became very slow, but gets
from that server are still as fast as always. The server and the client
are both on a local network with 10/100 nic. My ifconfig stats are as
follows

# ifconfig
eth1 Link encap:Ethernet HWaddr 00:50:04:E2:E3:C1
inet addr:192.168.1.10 Bcast:192.168.1.255
Mask:255.255.255.0
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:968943 errors:0 dropped:0 overruns:30320 frame:0
TX packets:714096 errors:0 dropped:0 overruns:0 carrier:1
collisions:0 txqueuelen:100
Interrupt:9 Base address:0xfc80

lo Link encap:Local Loopback
inet addr:127.0.0.1 Mask:255.0.0.0
UP LOOPBACK RUNNING MTU:3924 Metric:1
RX packets:220 errors:0 dropped:0 overruns:0 frame:0
TX packets:220 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0

I don't know what the overruns are an indication of but I have a feeling
that is where my problem is. I really don't think the problem is with
the ftp server but here is the version I am running
FTP server (Version wu-2.6.1-0.6x.21) ready.

This machine has a standard redhat 6.2 install with quite a few security
updates I did a while back.

Any help on solving this problem will be very welcome.

--
Chris W

"They that can give up essential liberty
to obtain a little temporary safety
deserve neither liberty nor safety."
-- Benjamin Franklin, 1759 Historical Review of Pennsylvania

Chris W wrote:

>I am running a wu-ftp server on redhat 6.2 on a 90mhz pentium with 72mb
>of ram. Recently sends to the FTP server became very slow, but gets
>from that server are still as fast as always. The server and the client
>are both on a local network with 10/100 nic. My ifconfig stats are as
>follows
>
># ifconfig
>eth1 Link encap:Ethernet HWaddr 00:50:04:E2:E3:C1
> inet addr:192.168.1.10 Bcast:192.168.1.255
>Mask:255.255.255.0
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:968943 errors:0 dropped:0 overruns:30320 frame:0
> TX packets:714096 errors:0 dropped:0 overruns:0 carrier:1
> collisions:0 txqueuelen:100
> Interrupt:9 Base address:0xfc80
>
>lo Link encap:Local Loopback
> inet addr:127.0.0.1 Mask:255.0.0.0
> UP LOOPBACK RUNNING MTU:3924 Metric:1
> RX packets:220 errors:0 dropped:0 overruns:0 frame:0
> TX packets:220 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:0
>
>I don't know what the overruns are an indication of but I have a feeling
>that is where my problem is. I really don't think the problem is with
>the ftp server but here is the version I am running
>FTP server (Version wu-2.6.1-0.6x.21) ready.
>
>This machine has a standard redhat 6.2 install with quite a few security
>updates I did a while back.
>
>Any help on solving this problem will be very welcome.
>
>--
>Chris W
>
>"They that can give up essential liberty
>to obtain a little temporary safety
>deserve neither liberty nor safety."
>-- Benjamin Franklin, 1759 Historical Review of Pennsylvania
>
>
>
>
>
Don't know if this is your problem, but a mismatch between full duplex
and half duplex config
of the NICs can cause this. In another case like this the culprit was
actually a sick card.

HTH

Michael

On Tue, 12 Aug 2003 11:36:08 -0500, Chris W wrote:

> I am running a wu-ftp server on redhat 6.2 on a 90mhz pentium with 72mb
> of ram. Recently sends to the FTP server became very slow, but gets
> from that server are still as fast as always. The server and the client
> are both on a local network with 10/100 nic. My ifconfig stats are as
> follows
>
> # ifconfig
> eth1 Link encap:Ethernet HWaddr 00:50:04:E2:E3:C1
> inet addr:192.168.1.10 Bcast:192.168.1.255
> Mask:255.255.255.0
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:968943 errors:0 dropped:0 overruns:30320 frame:0
> TX packets:714096 errors:0 dropped:0 overruns:0 carrier:1
> collisions:0 txqueuelen:100
> Interrupt:9 Base address:0xfc80
>
> lo Link encap:Local Loopback
> inet addr:127.0.0.1 Mask:255.0.0.0
> UP LOOPBACK RUNNING MTU:3924 Metric:1
> RX packets:220 errors:0 dropped:0 overruns:0 frame:0
> TX packets:220 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:0
>
> I don't know what the overruns are an indication of but I have a feeling
> that is where my problem is. I really don't think the problem is with
> the ftp server but here is the version I am running
> FTP server (Version wu-2.6.1-0.6x.21) ready.
>
> This machine has a standard redhat 6.2 install with quite a few security
> updates I did a while back.
>
> Any help on solving this problem will be very welcome.
>
> --

Pop a new NIC in it for a test.

Michael Krzepkowski wrote:

> Don't know if this is your problem, but a mismatch between full duplex
> and half duplex config
> of the NICs can cause this. In another case like this the culprit was
> actually a sick card.

on the win2k machine I am connecting to the ftp server from the duplex
setting is "hardware default" whatever that means, and I don't know how to
find or set the duplex mode on the linux machine

--
Chris W

"They that can give up essential liberty
to obtain a little temporary safety
deserve neither liberty nor safety."
-- Benjamin Franklin, 1759 Historical Review of Pennsylvania

>
> on the win2k machine I am connecting to the ftp server from the duplex
> setting is "hardware default" whatever that means, and I don't know how to
> find or set the duplex mode on the linux machine
>

Check the link lights on the card and whatever its connecting to on the
other end,. usually their is a high speed light and sometimes a full
duplex light.


> --
> Chris W
>
> "They that can give up essential liberty
> to obtain a little temporary safety
> deserve neither liberty nor safety."
> -- Benjamin Franklin, 1759 Historical Review of Pennsylvania

Frank Sweetser wrote:

> Chris W <(E-Mail Removed)> wrote:
> > Michael Krzepkowski wrote:
> >
> >> Don't know if this is your problem, but a mismatch between full duplex
> >> and half duplex config
> >> of the NICs can cause this. In another case like this the culprit was
> >> actually a sick card.
> >
> > on the win2k machine I am connecting to the ftp server from the duplex
> > setting is "hardware default" whatever that means, and I don't know how to
> > find or set the duplex mode on the linux machine
>
> The "hardware default" setting really means "I'm not telling". It will let the
> card set itself to whatever it likes, and give no indication of any mismatch
> problems. Set it to what it should be.

Both nics are 3com, the one in the windows machine is a 3C905B-tx the one in
the linux machine is a 3C905C. I also have an old 10 mbit generic no name NE2000
card in that machine that I'm not using. Just for the heck of it I hooked that
nic up to the switch to see what would happen. I used ftp to send and get a 100+
Mb files and the send transferred at 264.56Kbytes/sec while the get transferred at
668.82Kbytes/sec. I was also watching a network performance monitor and the send
graph is much more jagged than the receive but still not as jagged as with the 100
mbit card. I also changed the card in the windows machine to full duplex and then
half duplex with no change in the results. It also seems that the longer the
transfer goes on with the send, the more jagged the graph gets. One time the send
was
going along at its normally slow speed when about 80 MB into the transfer it
slowed down to a ridiculous crawl, I'm talking slow modem speeds. I have no ideas
at this point. The NE2000 card has no overruns like the 3com card does at this
point with about 300 mb sent and received. As a side note the NE2000 is eth0 and
the 3com is eth1. If there is no cable in the NE2000 I have to shut down eth0
before I can get a network connection on eth1. I have no idea why that is either.

--
Chris W

"They that can give up essential liberty
to obtain a little temporary safety
deserve neither liberty nor safety."
-- Benjamin Franklin, 1759 Historical Review of Pennsylvania

Frank Sweetser wrote:

> The "hardware default" setting really means "I'm not telling". It will let the
> card set itself to whatever it likes, and give no indication of any mismatch
> problems. Set it to what it should be.

Both nics are 3com, the one in the windows machine is a 3C905B-tx the one
inthe linux machine is a 3C905C. I also have an old 10 mbit generic no
name NE2000 card in that machine that I'm not using. Just for the heck of it
I hooked that nic up to the switch to see what would happen. I used ftp to
send and get a 100+ Mb file and the send transferred at 264.56Kbytes/sec
while the get transferred at 668.82Kbytes/sec. I was also watching a
network performance monitor and the send graph is much more jagged
than the receive but still not as jagged as with the 100 mbit card. I also
changed the card in the windows machine to full duplex and then half
duplex with no change in the results. It also seems that the longer the
transfer goes on the send the more jagged the graph gets. One time the
send was going along at its normally slow speed when about 80 MB into
the transfer it slowed down to a ridiculous crawl, I'm talking slow modem
speeds. I have no ideas at this point. The NE2000 card has no overruns
like the 3com card does at this point with about 300 mb sent and received.
As a side note the NE2000 is eth0 and the 3com is eth1. If there is no cable
in the NE2000 I have to shut down eth0 before I can get a network
connection on eth1. I have no idea why that is either.

Chris W

On Tue, 12 Aug 2003 11:36:08 -0500, Chris W <(E-Mail Removed)> wrote:
> I am running a wu-ftp server on redhat 6.2 on a 90mhz pentium with 72mb
> of ram. Recently sends to the FTP server became very slow, but gets
> from that server are still as fast as always. The server and the client
> are both on a local network with 10/100 nic. My ifconfig stats are as
> follows
>
> # ifconfig
> eth1 Link encap:Ethernet HWaddr 00:50:04:E2:E3:C1
> inet addr:192.168.1.10 Bcast:192.168.1.255
> Mask:255.255.255.0
> UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
> RX packets:968943 errors:0 dropped:0 overruns:30320 frame:0
> TX packets:714096 errors:0 dropped:0 overruns:0 carrier:1
> collisions:0 txqueuelen:100
> Interrupt:9 Base address:0xfc80
>
> lo Link encap:Local Loopback
> inet addr:127.0.0.1 Mask:255.0.0.0
> UP LOOPBACK RUNNING MTU:3924 Metric:1
> RX packets:220 errors:0 dropped:0 overruns:0 frame:0
> TX packets:220 errors:0 dropped:0 overruns:0 carrier:0
> collisions:0 txqueuelen:0
>
> I don't know what the overruns are an indication of but I have a feeling
> that is where my problem is. I really don't think the problem is with
> the ftp server but here is the version I am running
> FTP server (Version wu-2.6.1-0.6x.21) ready.
>
> This machine has a standard redhat 6.2 install with quite a few security
> updates I did a while back.
>
> Any help on solving this problem will be very welcome.

It could be a bad cable. If the RX pair is damaged at any point, that might
cause the errors you're seeing.

--
-- Skylar Thompson ((E-Mail Removed))
-- http://os2.dhs.org/~skylar/