Floris Bruynooghe
Floris Bruynooghe
We currently try to lookup both A and AAAA records in DNS for probes. But our probes, excluding https which handles it's own DNS, are split per protocol anyway. We...
There have been reliability issues with icmp probes. We've seen some cases of the pinger getting confused about what replies it is receiving, somehow having lost state. Maybe we should...
There are a whole bunch of flaky integration tests in the following files: - `iroh/tests/cli.rs` - `iroh/tests/provide.rs` I suspect all of these are related to filesystem latencies. We probably need...
Users are reporting slow thoughput on 10 Gbps links. While saturating that is a tough call we should aim to be close to saturating 1 Gbps
I'm not sure if the code itself is flaky or whether this is a race introduced in the filesystem use. https://github.com/n0-computer/iroh/actions/runs/7167046091/job/19512351881?pr=1874
Since https://github.com/n0-computer/iroh/pull/1858 we use hyper 1.0. Which means we currently have hyper duplicates in our dependency. This is a tracking issue to get all our dependencies upgraded: - https://github.com/dariusc93/rust-igd/issues/4 -...
If too many downloads are already in progress new requests should fail with a loadshed response.
A few crates did not have their minimal version set correctly. This bumps their versions. Check with: rm -f Cargo.lock cargo +nightly check -Z minimal-versions --all-features
IIUC the hickory DNS configuragion we use ends up with doing DNS requests using UDP with a 5 seconds timeout with two attempts, i.e. one retry. These retries are sequential,...
At the beginning of a connection we have observed it takes a while before we get a reasonable and stable Round-Trip-Time. We need to be a bit better at this:...