You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Doing a medium to large data transfer using the transfer example fails on main, with a timeout.
This fails even when running a local relay.
Steps to reproduce: you need 3 terminals, 1 sender, 1 receiver, 1 relay. Sender and receiver need to be set to relay only mode, which currently works by setting DEV_RELAY_ONLY=1 at compile time.
Relay: RUST_LOG=error cargo run --release -p iroh-relay --features server -- --dev
Sender: DEV_RELAY_ONLY=1 cargo run --release --example transfer --all-features -- provide --size=1g --relay-url http://127.0.0.1:3340
Receiver: DEV_RELAY_ONLY=1 cargo run --release --example transfer --all-features -- fetch --relay-url http://127.0.0.1:3340 <ticket_from_sender>
The transfer will slow down and then fail after a few megabytes.
Let's close this for now then. I handwavingly explain the issues as effective deadlocks or crawling to an extremely slow crawl of flows through the magic socket actors on the way to the relay server. Hence connection timeouts.
Doing a medium to large data transfer using the transfer example fails on main, with a timeout.
This fails even when running a local relay.
Steps to reproduce: you need 3 terminals, 1 sender, 1 receiver, 1 relay. Sender and receiver need to be set to relay only mode, which currently works by setting
DEV_RELAY_ONLY=1
at compile time.The transfer will slow down and then fail after a few megabytes.
For a detailed discussion, see https://discord.com/channels/949724860232392765/1324353518017974354
The issue seems to be fixed on #3077, so hopefully it will still be fixed when this code makes it to main.
The text was updated successfully, but these errors were encountered: