FC2 - Strange Ethernet Behavior

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hello everyone.

First, a big "thank you" to the Fedora team. Hats off, no pun. FC is impressing me more & more with each release. But...

I have a rather strange behavior going on for _ALL_ FC2 machines my company operates, and I'm hoping you can help me figure it out. This problem is **not** affecting any Windows-based hosts of any flavor, nor affecting other linux distros we run, such as Debian Woody and Sarge, all running on the same physical LAN, in the same physical segment. It's only affecting FC2 installations. This behavior is occuring **regardless of the NIC installed** - tulip-based, e100, 3com cards, you name it, doesn't make a difference. Because of this, I believe it's something deeper in the system, I figure, but I'll let you make that call. If it is, I have no idea what to do about it.

The behavior is this: our main connection is a T1. Top download speed, ie when no heavy traffic is going on, should hover somehwere around 150 K/sec, yes? That's been my experience. However, on **all** FC 2 machines we have, regardless of hardware, and regardless of running the original 2.6.5 kernel or the updated 2.6.7 kernel (yum update), download speed does something odd. It jumps up to about 180K/Sec, which causes alarms to go off at our ISP since that's above a T1's theo. limit, and then, shortly after 1.1 MB of file has downloaded, the connection speed plummets to less than old 28K dialup, ie, 2 K/Sec or less. It's still downloading mind you, but barely.

Doesn't matter what you're DL'ing, where it is, or how large it is as long as it's above 1MB. I guess that 1MB part is significant, as that's where the performance shift starts to happen. To rule out traffic, I cancelled the FC2 download of a given thing, say, webmin's tgz, and told a windows client to DL it from the same mirror (UMN). No problem. Then I did it on a Debian system. No problem. Then on FC2 again. Problem persists on FC2. Did all three at the same time - FC2 fails, the others split the avail bandwidth 50/50. Tried to do it via curl in the middle of the night; problem persists. Argh.

Because this problem is not affecting anything but the FC2 machines, and because all our FC2 machines have different networking hardware, I have to conclude it's something about FC2 itself. To strengthen this point, the machine from which I write this email, now running FC2 and having this problem, used to be running Mandrake 10 on the same hardware, just 2 days ago, with no troubles in networking.

I have no idea how this affecting traffic in the other direction. I hope it's not causing the same behavior for folks pulling stuff off of the FC2 machines. I'll have to test that.

So, anybody got any relevant ideas? Or am I destined to go hit Bugzilla for a fresh report?

Oh, btw... already played around with all the mii-tool options.  no effect.

John K., MCSE
(but rapidly converting to Linux!)



[Index of Archives]     [Current Fedora Users]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite News]     [Yosemite Photos]     [KDE Users]     [Fedora Tools]     [Fedora Docs]

  Powered by Linux