1540947 - Consider disabling HTTP/0.9 on Nightly

Status: NEW

(Core :: Networking, defect, P3)

Description

[Anne (:annevk)]

According to mt there is no usage on Nightly and Beta. Seems like it's time to start deprecating it and perhaps let folks on Nightly run into issues when they encounter it anyway.

Last attempt by Google: https://groups.google.com/a/chromium.org/d/msg/blink-dev/OdKnpLlvVUo/1EpFGVUjAwAJ

Comment 1

[Martin Thomson [:mt:]]

From our discussions:

Dragana Damjanovic there is 3.71M requests on beta in 10 days

Dragana Damjanovic https://telemetry.mozilla.org/new-pipeline/dist.html#!cumulative=0&end_date=2019-03-28&include_spill=0&keys=__none__!__none__!__none__&max_channel_version=beta%252F67&measure=HTTP_RESPONSE_VERSION&min_channel_version=null&processType=*&product=Firefox&sanitize=1&sort_by_value=0&sort_keys=submissions&start_date=2019-03-18&table=1&trim=1&use_submission_date=0

This is a very small proportion (0.01%) of our overall (of 1.18T), but it is still a number we can't ignore.

Comment 2

[Honza Bambas (:mayhemer)]

Do we recognize a 'true' H.9 response from a broken (out of framing) server responses? - if that is even possible...

Comment 3

[Dragana Damjanovic [:dragana]]

(In reply to Honza Bambas (:mayhemer) from comment #2)

Do we recognize a 'true' H.9 response from a broken (out of framing) server responses? - if that is even possible...

it is not possible. We do recognize 0.9 on a non-standards port, which are provably 'true' 0.9

https://telemetry.mozilla.org/new-pipeline/dist.html#!cumulative=0&end_date=2019-03-17&include_spill=0&keys=__none__!__none__!__none__&max_channel_version=nightly%252F67&measure=HTTP_09_INFO&min_channel_version=null&processType=*&product=Firefox&sanitize=1&sort_by_value=0&sort_keys=submissions&start_date=2019-01-28&table=0&trim=1&use_submission_date=0