Re: explanation of TCPLoss counters

[Selena Deckelmann]

On Wed, Sep 17, 2008 at 12:06 PM, Bernd Eckenfels <ecki@xxxxxxxxxxxx> wrote:
> In article <2b5e566d0809171014lb241f9am45250af2a366b128@xxxxxxxxxxxxxx> you wrote:
>> TCPLossUndo counter is being incremented when some errors occur.
>
> AFAIK this shows a reordering problem. (missing segments coming in late).

Ahhh!  I'm also seeing a single TCPDSACKRecv and a single TCPTimeout
at the same time.

Does TCPDSACKRecv mean that the SACK was dropped?

> Are you using multi path, bonding or strange switch topologies?

I am not sure, but I will look into it and if I am able to get that
information, I'll report back.

> The most reliable way to check out those extended TCP statistics is
> (unfortunatelly) the kernel source. But that most often requires expert know
> how on those algorithms (like selective acknowledgments, retransmits, reno,
> congestion control, undo window reduction, etc).

Right. Well, I appreciate your help.  I'm headed to the Linux Plumbers
Conference tomorrow, so perhaps I will have something to talk about :)

-selena

-- 
Selena Deckelmann
PDXPUG - http://pugs.postgresql.org/pdx
Me - http://www.chesnok.com/daily
--
To unsubscribe from this list: send the line "unsubscribe linux-net" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html