examples(iroh): allow to configure the receive window in the transfer example

[Frando]

This allows to configure the receive window in the transfer example. It can either be set by value, or by setting the expected RTT and then calculating the receive window for a desired max throughput of 12.5 MB/s (like in noq's default calculation).

For a relay-only transfer (with --relay-only) and the relay pinned to use1 (~100ms one-way latency from my laptop, so 400ms RTT between two endpoints on my laptop), increasing this from the default to 5MB (i.e. --receive-window 5Mor --receive-window-rtt 400) lets the transfer throughput go from ~600 KB/s to ~1.8 MB/s.

Description

Breaking Changes

Notes & open questions

Change checklist

• Self-review.
• Documentation updates following the style guide, if relevant.
• Tests if relevant.
• All breaking changes documented.
  • List all breaking changes in the above "Breaking Changes" section.
  • Open an issue or PR on any number0 repos that are affected by this breaking change. Give guidance on how the updates should be handled or do the actual updates themselves. The major ones are:
    • quic-rpc
    • iroh-gossip
    • iroh-blobs
    • dumbpipe
    • sendme

[github-actions]

Documentation for this PR has been generated and is available at: https://n0-computer.github.io/iroh/pr/4082/docs/iroh/

Last updated: 2026-04-08T08:19:38Z

[github-actions]

Netsim report & logs for this PR have been generated and is available at: LOGS
This report will remain available for 3 days.

Last updated for commit: 9380187

[flub]

We probably need to limit the number of features we throw into this example. Yes, it's a tool for our testing but also is this going to be generally useful enough to clutter the CLI?

Perhaps this is better off as a one-off thing to play with? Or perhaps the code needs to be structured differently so that it is easy to do this kind of tweaks locally if you need them?

Another concern is that tweaking the receive window without also tweaking the send window is also going to hurt at some point.

[flub]

Apparently github didn't think that counted as a review 🤷

Anyway, I need more convincing that we want to add this complexity to the transfer example :)

[Frando]

Anyway, I need more convincing that we want to add this complexity to the transfer example :)

My reasoning is:

• The transfer example is what we use to test connectivity and performance of iroh, and where we point people towards for testing themselves under varying conditions
• The default receive window size is far too low for high-latency conditions
• Making it tunable here allows people to test with higher sizes without having to write code themselves, to see if that is there issue

So I think that this argument makes sense here.

Another concern is that tweaking the receive window without also tweaking the send window is also going to hurt at some point.

Good point. The default is 8 * STREAM_RWND so I think I'd just do the same here.

[flub]

Do we no longer send folks to doctor? Anyway, you've done more maintaining of transfer than me, so your call. Though I do think it keeps getting more and more complex.

Do you plan on doing the send window as well or do you want an lgtm now?