Provide a try_bind_with_graceful_shutdown method to TlsServer. #717
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
…order to have fallible semantics rather than panics when a TLS configuration fails.
jxs
added
the
waiting-on-reviewer
bors bot
pushed a commit
to sigp/lighthouse
that referenced
this pull request
Oct 12, 2021
Currently, the beacon node has no ability to serve the HTTP API over TLS. Adding this functionality would be helpful for certain use cases, such as when you need a validator client to connect to a backup beacon node which is outside your local network, and the use of an SSH tunnel or reverse proxy would be inappropriate. ## Proposed Changes - Add three new CLI flags to the beacon node - `--http-enable-tls`: enables TLS - `--http-tls-cert`: to specify the path to the certificate file - `--http-tls-key`: to specify the path to the key file - Update the HTTP API to optionally use `warp`'s [`TlsServer`](https://docs.rs/warp/0.3.1/warp/struct.TlsServer.html) depending on the presence of the `--http-enable-tls` flag - Update tests and docs - Use a custom branch for `warp` to ensure proper error handling ## Additional Info Serving the API over TLS should currently be considered experimental. The reason for this is that it uses code from an [unmerged PR](seanmonstar/warp#717). This commit provides the `try_bind_with_graceful_shutdown` method to `warp`, which is helpful for controlling error flow when the TLS configuration is invalid (cert/key files don't exist, incorrect permissions, etc). I've implemented the same code in my [branch here](https://github.com/macladson/warp/tree/tls). Once the code has been reviewed and merged upstream into `warp`, we can remove the dependency on my branch and the feature can be considered more stable. Currently, the private key file must not be password-protected in order to be read into Lighthouse.
|
Is there any progress on this? Would love to see this merged. |
|
@seanmonstar Any thoughts on this one? I'm gonna close it if it's ignored any longer. Thanks. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes #680
Side notes:
Into<SocketAddr> + 'staticrather than just anInto<SocketAddr>, as well as returning aFuture<Output = ()> + 'staticrather than just aFuture<Output = ()>. There are no tests to show a difference in behavior in terms of what contexts a static lifetime is warranted. Removing all forms of+ 'staticand we compile all tests and they all pass. Likewise with thesignalparameters being+ Send + 'static...this doesn't seem necessary because hyper's with_graceful_shutdown does not requireSend + 'staticsemantics. Can we just remove all of these extra trait & lifetime bounds?try_binddoesn't return aResultbut instead returns nothing and just does a tracing event. It should behave likebindand calltry_bind_ephemeraland return aResult<impl Future<Output = ()>, crate::Error>Server::bindsay they return a Future but they don't, they wait on the future, but are anasync fnand so technically they do return a future, but one that awaits on the inner server future. It's kind of a confusing interface to passersby that do not understand the distinction even though they are technically the same. Should we re-write these methods to be regular functions and return an explicitFuturejust to be consistent?