Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 11 Oct 2024 18:14:11 +0200
From:      Michael Tuexen <michael.tuexen@lurchi.franken.de>
To:        Alan Somers <asomers@FreeBSD.org>
Cc:        FreeBSD Net <freebsd-net@freebsd.org>
Subject:   Re: How does the TCP measurement period work?
Message-ID:  <5F7D477B-A491-4851-890B-993B493D26B8@lurchi.franken.de>
In-Reply-To: <CAOtMX2jmXNvbEbO%2Bnjec-qGXcuPcrEr4qa-nrhdHjS8%2BrnutJA@mail.gmail.com>
References:  <CAOtMX2hKZy9omwHXLpKw42QwpGcUmTwLqSp=OWYYZ8cqOwwQ6w@mail.gmail.com> <7FB6F5E5-C6C0-4C18-B5FD-AEB18F0D5967@lurchi.franken.de> <CAOtMX2jmXNvbEbO%2Bnjec-qGXcuPcrEr4qa-nrhdHjS8%2BrnutJA@mail.gmail.com>

next in thread | previous in thread | raw e-mail | index | archive | help
> On 11. Oct 2024, at 14:55, Alan Somers <asomers@FreeBSD.org> wrote:
>=20
> On Fri, Oct 11, 2024 at 1:05=E2=80=AFAM Michael Tuexen
> <Michael.Tuexen@lurchi.franken.de> wrote:
>>=20
>>> On 11. Oct 2024, at 01:07, Alan Somers <asomers@freebsd.org> wrote:
>>>=20
>>> Can somebody please explain to me how the TCP measurement period
>>> works?  When does h_ertt decide to take a new measurement?
>>>=20
>>> Motivation:
>>> I recently saw a long-distance connection that should've been =
capable
>>> of 80+ MBps suddenly drop to < 1 MBps.  Subsequent analysis of the
>>> pcap file showed that while the typical RTT was 16.5 ms, there were =
a
>>> few spikes as high as 380ms that coincided with the drop in
>>> throughput.  The surprising part was that even though RTT returned =
to
>>> a good value, the throughput stayed low for the entire remaining
>>> transfer, which lasted 750s.  I would've expected throughput to
>>> recover once RTT did.  My theory is that h_ertt never made a new
>>> measurement.  However, I cannot reproduce the problem using dummynet
>>> on a local VM.  With dummynet, as soon as I return the RTT to =
normal,
>>> the throughput quickly recovers, as one would expect.
>> Which TCP stack and which CC module did you use? Which version of =
FreeBSD?
>=20
> I was using the regular freebsd TCP stack with cc_chd.  The production
I'm not sure how well debugged chd is...
> system was running FreeBSD 14.1, but my attempts to recreate the
> situation using dummynet used 15.0-CURRENT.
Please report, if you are able to reproduce it. In that case some
BBLogs might be useful.

Best regards
Michael




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?5F7D477B-A491-4851-890B-993B493D26B8>