Any type of bittorrent downloads kills linux network
Linux - NetworkingThis forum is for any issue related to networks or networking.
Routing, network cards, OSI, etc. Anything is fair game.
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
Any type of bittorrent downloads kills linux network
If I start any kind of torrent download all network traffic on my box dies and I get the following error in /var/log/messages:
Aug 22 16:47:58 kernel: NETDEV WATCHDOG: eth0: transmit timed out
Aug 22 16:47:58 kernel: tg3: eth0: transmit timed out, resetting
Aug 22 16:47:58 kernel: tg3: tg3_stop_block timed out, ofs=1400 enable_bit=2
Aug 22 16:47:58 kernel: tg3: tg3_stop_block timed out, ofs=c00 enable_bit=2
Aug 22 16:47:58 kernel: tg3: eth0: Link is down.
Aug 22 16:48:02 kernel: tg3: eth0: Link is up at 1000 Mbps, full duplex.
Aug 22 16:48:02 kernel: tg3: eth0: Flow control is on for TX and on for RX.
Sometimes the link doesn't come back up and I need to restart the service.
Running on a shuttle box SD31P
2.6.17-1.2174_FC5smp #1 SMP Tue Aug 8 16:00:39 EDT 2006 i686 i686 i386 GNU/Linux
I'm not sure if its just a torrent thing or a network thing. I can run torrents fine on a windows box that is on the same lan going out through the same router. I can download huge amounts of data through ftp just fine. Its only with torrents. I thought initially it was a bad python install so I grabbed a C based torrent client and it failed also.
Is there some kind of limit that can be adjusted in the kernel relating to maximum number of open connections? Any thoughts or suggestions?
Distribution: Red Hat Enterprise Linux, Debian & Ubuntu
Posts: 92
Rep:
Whenever I run into this problem, it ends up being an issue with the NIC-to-Router communication. The router will kill the Ethernet connection due to a bug-- the memory cache on the router might fill up, or there is some other bug.
When this happens to me, I reduce the maximum-download speed by 5% or so. This has always fixed the problem for me.
Yeah, I tried doing that. What makes me think its not the router is because I have a windows machine sitting on the same router that doesn't have problems and the messages on the linux box. I dropped the upload speed, but download has always been unlimited (same as the windows box). I'll try dropping the download speed some and see what happens.
I dropped the download speed and attempted another torrent. About 30 seconds into the torrent starting I lose connection with my box, it didn't even have a chance to get anywhere near max download speed. Now I'm just waiting for the network to reset itself (if it even does that, sometimes I have to go to the terminal and reset it).
Was able to get back in, again the same is in /var/log/messages:
Aug 23 10:15:06 kernel: NETDEV WATCHDOG: eth0: transmit timed out
Aug 23 10:15:06 kernel: tg3: eth0: transmit timed out, resetting
Aug 23 10:15:06 kernel: tg3: tg3_stop_block timed out, ofs=2c00 enable_bit=2
Aug 23 10:15:06 kernel: tg3: tg3_stop_block timed out, ofs=1400 enable_bit=2
Aug 23 10:15:06 kernel: tg3: tg3_stop_block timed out, ofs=c00 enable_bit=2
Aug 23 10:15:06 kernel: tg3: eth0: Link is down.
Aug 23 10:15:09 kernel: tg3: eth0: Link is up at 1000 Mbps, full duplex.
Aug 23 10:15:09 kernel: tg3: eth0: Flow control is on for TX and on for RX.
Another update. I found a newsgroup posting from june with someone talking about exactly what I was seeing. Someone recommended they issue a:
ethtool -K eth0 tso off
Command and see if that helps. It seemed to help so far. The posts seem to imply that it is a kernel problem with the tg3 driver.
Where can I put this so that when I reboot the box it happens automatically?
I had the same problem, but a little bit of Googling revealed that several others have had the same problem and solved it by turning of TSO on you network card:
ethtool -K eth0 tso off
tso offloads some of the segmentation work from the OS to your network card, but it seems buggy. I turned it off and <knock on wood> haven't had a problem since.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.