Network dies under file transfer...
Greetings! If I transfer 150MB+ of files from Windows 98 machine to my file server using Samba, the network on the file server will occasionally die and a reboot is needed to restart the network. I always thought this was a Windows/Samba problem. When I transfer 150MB+ using FTP, the network will most definitely die. Anybody know what this could be? I'm using a SuSE 6.3 distribution, the standard 2.2.18 kernel with the 0.90 RAID and 3.5.29 ReiserFS patches, on a Cyrix M2 PR-333MHz, 64MB, MVP4 motherboard, and a Netgear FA310TX (tulip) NIC. Most of the network options are off in the standard kernel, and I only have a few turned on that are directly relevant to this particular machine. (It's a file server, not a router.) Thanks! Christopher Reimer
Are there any error messages under /var/log/messages? On Saturday 27 January 2001 11:40 am, Christopher D. Reimer wrote:
Greetings!
If I transfer 150MB+ of files from Windows 98 machine to my file server using Samba, the network on the file server will occasionally die and a reboot is needed to restart the network. I always thought this was a Windows/Samba problem. When I transfer 150MB+ using FTP, the network will most definitely die. Anybody know what this could be?
I'm using a SuSE 6.3 distribution, the standard 2.2.18 kernel with the 0.90 RAID and 3.5.29 ReiserFS patches, on a Cyrix M2 PR-333MHz, 64MB, MVP4 motherboard, and a Netgear FA310TX (tulip) NIC. Most of the network options are off in the standard kernel, and I only have a few turned on that are directly relevant to this particular machine. (It's a file server, not a router.)
Thanks!
Christopher Reimer
None. :( At 12:22 PM 1/27/2001, Matthew wrote:
Are there any error messages under /var/log/messages?
On Saturday 27 January 2001 11:40 am, Christopher D. Reimer wrote:
Greetings!
If I transfer 150MB+ of files from Windows 98 machine to my file server using Samba, the network on the file server will occasionally die and a reboot is needed to restart the network. I always thought this was a Windows/Samba problem. When I transfer 150MB+ using FTP, the network will most definitely die. Anybody know what this could be?
I'm using a SuSE 6.3 distribution, the standard 2.2.18 kernel with the 0.90 RAID and 3.5.29 ReiserFS patches, on a Cyrix M2 PR-333MHz, 64MB, MVP4 motherboard, and a Netgear FA310TX (tulip) NIC. Most of the network options are off in the standard kernel, and I only have a few turned on that are directly relevant to this particular machine. (It's a file server, not a router.)
Thanks!
Christopher Reimer
I had a similar problem when transferring files between two systems, one with a Netgear FA310TX and the other with a 3C905B. After experimenting a bit, I discovered that if I could get the file transfers to continue without having to reboot or reenter any FTP commands by entering the following commands on the system with the Netgear NIC: # rcnetwork stop; sleep 1; rcnetwork start # rcroute start; rcfirewall start Jim Cunning On Sat, 27 Jan 2001, Christopher D. Reimer wrote:
None. :(
At 12:22 PM 1/27/2001, Matthew wrote:
Are there any error messages under /var/log/messages?
On Saturday 27 January 2001 11:40 am, Christopher D. Reimer wrote:
Greetings!
If I transfer 150MB+ of files from Windows 98 machine to my file server using Samba, the network on the file server will occasionally die and a reboot is needed to restart the network. I always thought this was a Windows/Samba problem. When I transfer 150MB+ using FTP, the network will most definitely die. Anybody know what this could be?
I'm using a SuSE 6.3 distribution, the standard 2.2.18 kernel with the 0.90 RAID and 3.5.29 ReiserFS patches, on a Cyrix M2 PR-333MHz, 64MB, MVP4 motherboard, and a Netgear FA310TX (tulip) NIC. Most of the network options are off in the standard kernel, and I only have a few turned on that are directly relevant to this particular machine. (It's a file server, not a router.)
At 01:06 PM 1/27/2001, jcunning@cts.com wrote:
I had a similar problem when transferring files between two systems, one with a Netgear FA310TX and the other with a 3C905B. After experimenting a bit, I discovered that if I could get the file transfers to continue without having to reboot or reenter any FTP commands by entering the following commands on the system with the Netgear NIC:
# rcnetwork stop; sleep 1; rcnetwork start # rcroute start; rcfirewall start Jim Cunning
It worked for a little while and then it failed again. Thanks! Christopher Reimer
I forgot to say that I had to do this multiple times during a multi-GB FTP transfer. Not fun, but at least I didn't have to reboot or restart the transfer. Jim On Sat, 27 Jan 2001, Christopher D. Reimer wrote:
At 01:06 PM 1/27/2001, jcunning@cts.com wrote:
I had a similar problem when transferring files between two systems, one with a Netgear FA310TX and the other with a 3C905B. After experimenting a bit, I discovered that if I could get the file transfers to continue without having to reboot or reenter any FTP commands by entering the following commands on the system with the Netgear NIC:
# rcnetwork stop; sleep 1; rcnetwork start # rcroute start; rcfirewall start Jim Cunning
It worked for a little while and then it failed again.
Thanks!
Christopher Reimer
-- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq
Have you tried Becker's latest drivers? You can get a source RPM that will update all the modules. Go to www.scyld.com. That may help... Matt On Saturday 27 January 2001 01:52 pm, Christopher D. Reimer wrote:
At 01:06 PM 1/27/2001, jcunning@cts.com wrote:
I had a similar problem when transferring files between two systems, one with a Netgear FA310TX and the other with a 3C905B. After experimenting a bit, I discovered that if I could get the file transfers to continue without having to reboot or reenter any FTP commands by entering the following commands on the system with the Netgear NIC:
# rcnetwork stop; sleep 1; rcnetwork start # rcroute start; rcfirewall start Jim Cunning
It worked for a little while and then it failed again.
Thanks!
Christopher Reimer
At 05:36 PM 1/27/2001, Matthew wrote:
Have you tried Becker's latest drivers? You can get a source RPM that will update all the modules.
Go to www.scyld.com.
That may help...
Matt
I'm having trouble trying to compile the driver, either as modules or kernel linked. Apparently, the drivers are supposed to be compiled against Red Hat linux. I'm using SuSE 6.3 distribution with the standard 2.2.18 kernel. Thanks! Christopher Reimer
They are not compiled against any distrobution that I know of and are not specially for Redhat Linux. After what Linus said about RH 7.0 I doubt he will use that. ftp://ftp.scyld.com/pub/network/netdriver-2.1.src.rpm I got them on by downloading them and then issuing the "rpm --rebuild netdriver-2.1.src.rpm" Then go to the /usr/src/packages/RPMS directory and you will see an RPM you can install issuing the "rpm --force -i -v netdriver-2.1.i386.rpm" You will have to reboot afterwoods. Check dmesg, or ctrl-alt-F10 and you should a date that is 2001 or 2000. Depending on what card you have be careful changing options, tried it on the Tulip driver and had a segfault on eth0, that hung whole boot process. Although it could be the card... Matt On Saturday 27 January 2001 11:47 pm, Christopher D. Reimer wrote:
At 05:36 PM 1/27/2001, Matthew wrote:
Have you tried Becker's latest drivers? You can get a source RPM that will update all the modules.
Go to www.scyld.com.
That may help...
Matt
I'm having trouble trying to compile the driver, either as modules or kernel linked. Apparently, the drivers are supposed to be compiled against Red Hat linux. I'm using SuSE 6.3 distribution with the standard 2.2.18 kernel.
Thanks!
Christopher Reimer
At 12:14 AM 1/28/2001, Matthew wrote:
They are not compiled against any distrobution that I know of and are not specially for Redhat Linux. After what Linus said about RH 7.0 I doubt he will use that.
ftp://ftp.scyld.com/pub/network/netdriver-2.1.src.rpm
I got them on by downloading them and then issuing the "rpm --rebuild netdriver-2.1.src.rpm"
I tried that... and it didn't work because it was complaining of unresolved symbols. I had better success with compiling the individual drivers: gcc -I/usr/src/linux/include -DMODULE -D__KERNEL__ -O6 -c tulip.c gcc -I/usr/src/linux/include -DMODULE -D__KERNEL__ -O6 -c pci-scan.c I had to include the -I option to successfully compile the drivers. insmod pci-scan.o insmod tulip.o These drivers are blazingly fast... I was getting a consistent 395 kilobytes/sec on my 10Mb network. Unfortunately, the network still dies after transferring 50% of a 1GB batch transfer. (This is significantly more than what I was able to do with the older drivers.) Thanks! Christopher Reimer
Hmmmm....... Always dieing at 50%? Have you tried a 500MB transfer? Tried different cables? This is a difficult one to trace to say the least..... Review everything you have done so far, might be something else we can give a shot on. Matt On Sunday 28 January 2001 12:33 pm, Christopher D. Reimer wrote:
At 12:14 AM 1/28/2001, Matthew wrote:
They are not compiled against any distrobution that I know of and are not specially for Redhat Linux. After what Linus said about RH 7.0 I doubt he will use that.
ftp://ftp.scyld.com/pub/network/netdriver-2.1.src.rpm
I got them on by downloading them and then issuing the "rpm --rebuild netdriver-2.1.src.rpm"
I tried that... and it didn't work because it was complaining of unresolved symbols. I had better success with compiling the individual drivers:
gcc -I/usr/src/linux/include -DMODULE -D__KERNEL__ -O6 -c tulip.c gcc -I/usr/src/linux/include -DMODULE -D__KERNEL__ -O6 -c pci-scan.c
I had to include the -I option to successfully compile the drivers.
insmod pci-scan.o insmod tulip.o
These drivers are blazingly fast... I was getting a consistent 395 kilobytes/sec on my 10Mb network.
Unfortunately, the network still dies after transferring 50% of a 1GB batch transfer. (This is significantly more than what I was able to do with the older drivers.)
Thanks!
Christopher Reimer
At 01:37 PM 1/28/2001, Matthew wrote:
Hmmmm.......
Always dieing at 50%? Have you tried a 500MB transfer? Tried different cables?
This is a difficult one to trace to say the least.....
Review everything you have done so far, might be something else we can give a shot on.
OK... I got my kernel properly configured. (Forgot to uncomment the INSTALL=/boot in the kernel Makefile... DOH!) The new tulip driver is correctly being loaded as a module, with "tulip debug=2 options=0" in /etc/modules.conf. I get the following error message three times in /var/log/messages before the network dies: Oversized Ethernet frame spanned multiple buffers, status 08018192! Now what? :) Christopher Reimer
Hey you are getting an error message! Thats something at least :-). I have little idea what it means though.....Its probably outside of most peoples realm here. Try subscribing to the tulip mailing list on scyld.com. Only takes a few minutes to setup, but someone may know precisely the causes of this problem. Who is the manufacturer of the card? Matt On Sunday 28 January 2001 04:46 pm, Christopher D. Reimer wrote:
At 01:37 PM 1/28/2001, Matthew wrote:
Hmmmm.......
Always dieing at 50%? Have you tried a 500MB transfer? Tried different cables?
This is a difficult one to trace to say the least.....
Review everything you have done so far, might be something else we can give a shot on.
OK... I got my kernel properly configured. (Forgot to uncomment the INSTALL=/boot in the kernel Makefile... DOH!) The new tulip driver is correctly being loaded as a module, with "tulip debug=2 options=0" in /etc/modules.conf. I get the following error message three times in /var/log/messages before the network dies:
Oversized Ethernet frame spanned multiple buffers, status 08018192!
Now what? :)
Christopher Reimer
At 06:31 PM 1/28/2001, Matthew wrote:
Hey you are getting an error message! Thats something at least :-).
I have little idea what it means though.....Its probably outside of most peoples realm here.
Try subscribing to the tulip mailing list on scyld.com. Only takes a few minutes to setup, but someone may know precisely the causes of this problem.
Who is the manufacturer of the card?
It's a Netgear FA310TX. I read through parts of the Scyld tulip list and discovered that this is a known bug with some recent versions of the tulip chipset. Basically, under heavy traffic that generates frame collisions, the chipset gets "confuse" and writes garbage to some frames. This causes the error message to happen, and, subsequently, shut downs the NIC. This is a common problem if the tulip NIC is not set to "auto-negotiation" and it's linked to a 10baseT hub. It's supposed to be less common if you are using a 100baseT hub. I verified that my file server and windows box were set correctly (they both have a FA310TX), I was able to transfer 95% of my files over before the network dropped out. I decided to get a new network card with a different chipset to eliminate the random crashes. I already spent several weeks determining that this wasn't a Windows/Samba/RAID/ReiserFS/kernel problem. I just subscribed to the tulip list, and will probably be on another Scyld list when I get the new card. Thanks for the help! Christopher Reimer
No problem! Wish I had the answer for you right away though. I am going to replace my NIC's too, with more expensive ones. No point getting cheap ones that end up having to be constantly replaced...Already building up a pile myself... If you were just using Windows you may not have ever known, or got this far. I do not trust anything about Windows error messages and or driver status in case they are lying. Did you by any chance recieve any weird FETCHMAIL-DAEMON messages? Trying to work out whether or not this is a security breach on my machine, or whether its something else. Matt On Sunday 28 January 2001 07:17 pm, Christopher D. Reimer wrote:
At 06:31 PM 1/28/2001, Matthew wrote:
Hey you are getting an error message! Thats something at least :-).
I have little idea what it means though.....Its probably outside of most peoples realm here.
Try subscribing to the tulip mailing list on scyld.com. Only takes a few minutes to setup, but someone may know precisely the causes of this problem.
Who is the manufacturer of the card?
It's a Netgear FA310TX. I read through parts of the Scyld tulip list and discovered that this is a known bug with some recent versions of the tulip chipset. Basically, under heavy traffic that generates frame collisions, the chipset gets "confuse" and writes garbage to some frames. This causes the error message to happen, and, subsequently, shut downs the NIC. This is a common problem if the tulip NIC is not set to "auto-negotiation" and it's linked to a 10baseT hub. It's supposed to be less common if you are using a 100baseT hub. I verified that my file server and windows box were set correctly (they both have a FA310TX), I was able to transfer 95% of my files over before the network dropped out.
I decided to get a new network card with a different chipset to eliminate the random crashes. I already spent several weeks determining that this wasn't a Windows/Samba/RAID/ReiserFS/kernel problem. I just subscribed to the tulip list, and will probably be on another Scyld list when I get the new card. Thanks for the help!
Christopher Reimer
At 07:52 PM 1/28/2001, Matthew wrote:
Did you by any chance recieve any weird FETCHMAIL-DAEMON messages? Trying to work out whether or not this is a security breach on my machine, or whether its something else.
Someone else's machine is having a problem. It's nothing to worry about. Christopher Reimer
Another option is to search and subscribe to the relevant modules mailing list. See if they can give any hints (I just joined the Tulip mailing list...) Matt On Saturday 27 January 2001 11:47 pm, Christopher D. Reimer wrote:
At 05:36 PM 1/27/2001, Matthew wrote:
Have you tried Becker's latest drivers? You can get a source RPM that will update all the modules.
Go to www.scyld.com.
That may help...
Matt
I'm having trouble trying to compile the driver, either as modules or kernel linked. Apparently, the drivers are supposed to be compiled against Red Hat linux. I'm using SuSE 6.3 distribution with the standard 2.2.18 kernel.
Thanks!
Christopher Reimer
Sounds to me like you might have a duplex mis-match. I've seen this elsewhere where one end is talking full-duplex & the other is half-duplex. The network is up and seems ok for pings, etc, but FTP file transfers take so hideously long that you might as well have put it on tape and carried it over to its destination. A slow FTP is a dead givaway for this. It may not be your problem, but check it out anyway. John -----Original Message----- From: Christopher D. Reimer [mailto:creimer@rahul.net] Sent: 27 January 2001 19:41 To: suse-linux-e@ns2.SuSE.com Subject: [SLE] Network dies under file transfer... Greetings! If I transfer 150MB+ of files from Windows 98 machine to my file server using Samba, the network on the file server will occasionally die and a reboot is needed to restart the network. I always thought this was a Windows/Samba problem. When I transfer 150MB+ using FTP, the network will most definitely die. Anybody know what this could be? I'm using a SuSE 6.3 distribution, the standard 2.2.18 kernel with the 0.90 RAID and 3.5.29 ReiserFS patches, on a Cyrix M2 PR-333MHz, 64MB, MVP4 motherboard, and a Netgear FA310TX (tulip) NIC. Most of the network options are off in the standard kernel, and I only have a few turned on that are directly relevant to this particular machine. (It's a file server, not a router.) Thanks! Christopher Reimer -- To unsubscribe send e-mail to suse-linux-e-unsubscribe@suse.com For additional commands send e-mail to suse-linux-e-help@suse.com Also check the FAQ at http://www.suse.com/support/faq
participants (4)
-
Christopher D. Reimer -
jcunning＠cts.com -
john -
Matthew