Jump to main content

Nicolas Hoizey

  • articles
  • billets
  • links
  • notes
  • talks
  • archives
  • about

Note from 3 July 2020

  • Nicolas Hoizey
  • 3 July 2020
  • HTTP, WebPerf
  • 11 reactions

@bagder hi Daniel, reading https://daniel.haxx.se/blog/2016/08/18/http2-connection-coalescing/ (yes, I'm late 😅)

If DNS B.example.com returns 192.168.0.2 and 192.168.0.3, is TLS really enough to be sure it also works on 192.168.0.1?

BTW, there's a typo near the end: "both IPv4 and IPv4 addresses".

11 reactions

11 replies

  1. Nicolas Hoizey avatar Nicolas Hoizey
    BTW, there's a typo near the end: "both IPv4 and IPv4 addresses".
    • 3 July 2020, 14:48
    • Source
  2. Daniel Stenberg avatar Daniel Stenberg
    I'm not sure what you're asking. Are you asking if my explanation is correct of how Firefox does connection coalescing?
    • 3 July 2020, 14:51
    • Source
  3. Nicolas Hoizey avatar Nicolas Hoizey
    I'm not questioning your explanation at all.

    I don't understand how Firefox can make sure B can answer on the same IP as A when DNS doesn't list it.
    • 3 July 2020, 15:05
    • Source
  4. Daniel Stenberg avatar Daniel Stenberg
    that's not what it says, it says it can reuse/coalesce the connection for A with B if DNS and the certs have overlap
    • 3 July 2020, 15:30
    • Source
  5. Nicolas Hoizey avatar Nicolas Hoizey
    As I understand it:
    - A is on 1 and 2
    - Firefox connects to 1
    - TLS says * are the same
    - B is on 2 and 3
    - 2 and 3 are in *, so 1 (which is also) is reused

    That's what I understand in "It will reuse […] A and ask for host B’s content over that single shared connection"
    • 3 July 2020, 15:57
    • Source
  6. Daniel Stenberg avatar Daniel Stenberg
    yes, that seems accurate
    • 3 July 2020, 15:57
    • Source
  7. Nicolas Hoizey avatar Nicolas Hoizey
    Ok, so I wonder how it reacts if B is not reachable with 192.168.0.1 … 🤷‍♂️
    • 3 July 2020, 17:53
    • Source
  8. Daniel Stenberg avatar Daniel Stenberg
    I'm pretty sure that doesn't matter. The point is that the site admins have specified that A and B overlap.
    • 3 July 2020, 17:56
    • Source
  9. Nicolas Hoizey avatar Nicolas Hoizey
    Ok, thanks.

    Maybe I should let go… 😉
    • 3 July 2020, 17:57
    • Source
  10. Daniel Stenberg avatar Daniel Stenberg
    you don't have to, but if you want to argue that they're doing it wrong you should talk to Firefox engineers, not me...
    • 3 July 2020, 17:59
    • Source
  11. Nicolas Hoizey avatar Nicolas Hoizey
    I really do not want to argue anything, just to understand how/why it works.
    • 3 July 2020, 18:00
    • Source
  • Older: Note from 2 July 2020
  • Newer: Note from 3 July 2020

Related contents with similar topics

  1. screenshot of Caching Header Best Practices

    Simon Hearne

    Caching Header Best Practices

    • Nicolas Hoizey
    • 14 October 2022
    • WebPerf, HTTP
    • 4 reactions

    Client-side caching is a key technique to improving front-end speed and user experience. Whilst it may appear complex and risky, investing the time…

  2. Note from 15 March 2022

    • Nicolas Hoizey
    • 15 March 2022
    • Cloudflare, Fastly, HTTP, WebPerf
    • 3 reactions

    TIL: #Cloudflare doesn't support the Vary HTTP header, which means the origin server can't do any content negotiation, for example send WebP or AVIF for a JPEG request…

    https://developers.cloudflare.com/cache/about/cache-control/#other

    #Fastly supports it, just saying… 😅

If you want to share an error or suggest an enhancement of this content, please edit the source on GitHub.

© Nicolas Hoizey

Built with Eleventy