With the introduction of the DNS cache for Clients in #10533, we now enable a behaviour where we don't necessarily need to establish a connection to a Gateway to resolve a DNS query if we still have a valid entry in the DNS cache. In particular, the proptests discovered that: - a DNS query for an upstream resolver - which happens to be a resource - and has a valid entry in the DNS cache - but (no longer) a connection to the corresponding Gateway will now serve the cached DNS records instead of establishing a new connection to the Gateway. As a result, the site status which we assert in the proptests remains in "unknown" instead of the expected "online". Modelling the caching behaviour in the tests is rather tedious. To avoid that, we set the TTL of all simulated upstream DNS responses to 1 which effectively bypasses the cache. Whilst not an ideal solution, it ensures that CI is consistently green without flaky tests. The DNS cache itself is already unit-tested.
Rust development guide
Firezone uses Rust for all data plane components. This directory contains the Linux and Windows clients, and low-level networking implementations related to STUN/TURN.
We target the last stable release of Rust using rust-toolchain.toml.
If you are using rustup, that is automatically handled for you.
Otherwise, ensure you have the latest stable version of Rust installed.
Reading Client logs
The Client logs are written as JSONL for machine-readability.
To make them more human-friendly, pipe them through jq like this:
cd path/to/logs # e.g. `$HOME/.cache/dev.firezone.client/data/logs` on Linux
cat *.log | jq -r '"\(.time) \(.severity) \(.message)"'
Resulting in, e.g.
2024-04-01T18:25:47.237661392Z INFO started log
2024-04-01T18:25:47.238193266Z INFO GIT_VERSION = 1.0.0-pre.11-35-gcc0d43531
2024-04-01T18:25:48.295243016Z INFO No token / actor_name on disk, starting in signed-out state
2024-04-01T18:25:48.295360641Z INFO null
Benchmarking on Linux
The recommended way for benchmarking any of the Rust components is Linux' perf utility.
For example, to attach to a running application, do:
- Ensure the binary you are profiling is compiled with the
releaseprofile. sudo perf record -g --freq 10000 --pid $(pgrep <your-binary>).- Run the speed test or whatever load-inducing task you want to measure.
sudo perf script > profile.perf- Open profiler.firefox.com and load
profile.perf
Instead of attaching to a process with --pid, you can also specify the path to executable directly.
That is useful if you want to capture perf data for a test or a micro-benchmark.