Any luck replicating this? Does anyone else have any ideas?
--- Henrique de Moraes Holschuh <[EMAIL PROTECTED]> wrote:
> On Sun, 12 Feb 2006, Mark Flocco wrote:
> > session. However, during times of high traffic (~8Mbps), the RTTs
>
> 8Mbits/s or 8Mbytes/s ? Anyway neither is much of a high tr
> 8Mbits/s or 8Mbytes/s ?
8Mbits/s. That's at peak, as well.
>
> I wouldn't call that stress, unless your switch is really awful.
I'm seeing this across two different switches - a Cisco 3550 and a
Cisco 2912. I'm also seeing it in 100Mbps and 1000Mbps modes (2912
doesn't support Gbit)
> an
On Sun, 12 Feb 2006, Mark Flocco wrote:
> session. However, during times of high traffic (~8Mbps), the RTTs
8Mbits/s or 8Mbytes/s ? Anyway neither is much of a high traffic for an
e1000 on a modern host, unless you're using a very, very low MTU (and maybe
not even then).
> increase, showing stre
> And if you do that from a non e1000 host *to* an e1000 host?
Same result - variable times.
>
> Also, do you have traffic in the e1000 other than the ping? Heavy
> traffic?
> Light traffic? Is NAPI enabled in the kernel?
The ping I pasted was with zero other traffic other than an ssh
sessio
On Sun, 12 Feb 2006, Mark Flocco wrote:
> servers. I'm seeing variable round trip times when I send pings to any
> host from a host with the Intel Pro/1000 MT. Meaning, I don't have to
And if you do that from a non e1000 host *to* an e1000 host?
Also, do you have traffic in the e1000 other than
I'm currently running Debian sarge with the e1000 driver on a few
servers. I'm seeing variable round trip times when I send pings to any
host from a host with the Intel Pro/1000 MT. Meaning, I don't have to
ping between these hosts to get the variable RTTs. Comparitively, on
two other servers wh
6 matches
Mail list logo
