refactor(iroh-net): call-me-maybe improvements (no more tasks for queue, better logic on recv) #1752
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
dignifiedquire
force-pushed
the
iroh-net-refactor-netcheck-fix
branch
from
7d5f987 to
d8775ab
Compare
Frando
force-pushed
the
iroh-net-refactor-netcheck-fix-cmm-queue
branch
from
e3fdc48 to
70380c0
Compare
Frando
changed the title
This was referenced Oct 27, 2023
iroh-net/src/magicsock.rs
Outdated
| log_endpoint_change(&eps); | ||
| if let Some(ref cb) = self.inner.on_endpoints { | ||
| cb(&eps[..]); | ||
| } | ||
| for (public_key, region_id) in self.inner.pending_call_me_maybes.lock().drain() { | ||
| self.inner | ||
| .send_or_queue_call_me_maybe(region_id, public_key); |
There was a problem hiding this comment.
shouldn't this force sending instead of potentially requeuing? not sure in either direction
There was a problem hiding this comment.
Yea it won't requeue because endpoints.set was just called which means endpoints.is_fresh will always be true.
We could add a separate send_call_me_maybe method for clarity though.
There was a problem hiding this comment.
I think that would be good, to make the intent clear (given the complexity of this code.. )
There was a problem hiding this comment.
Changed it to separate functions.
dignifiedquire
approved these changes
Oct 27, 2023
github-merge-queue bot
pushed a commit
that referenced
this pull request
Oct 27, 2023
## Description In a longer debugging session with @dignifiedquire we found out that `wait_idle` will *always* wait on a close timer for a few 100 ms before actually completing. If application layer protocols always take care of finishing their work (which we do) this is not needed at all. Also the magicsock will very likely stay around long enough to still send out the close frame. This together with #1752 brings the runtime of the `derp_connect_loop` test down from 15s to 4s on my machine! First round is 2-3s (due to bugs in netcheck making it too slow likely) and subsequent round are all >500ms! ## Notes & open questions <!-- Any notes, remarks or open questions you have to make about the PR. --> ## Change checklist - [ ] Self-review. - [ ] Documentation updates if relevant. - [ ] Tests if relevant.
Frando
force-pushed
the
iroh-net-refactor-netcheck-fix-cmm-queue
branch
from
70380c0 to
a337405
Compare
dignifiedquire
approved these changes
Oct 28, 2023
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.
Description
Removes the task spawning for queued call-me-maybes. Instead just keep them in a hashmap. This is possible now because sending call-me-maybes is sync anyway. We will lose them if the derp channel is full, which might be a problem, but that is already a problem on main.
Additionally removes
ActorMessage::ReStunand instead moves theEndpointUpdateStateontoInner. One less thing to go over the main channel - not needed, because theEndpointUpdateStatealready is a concurrent data structure (with a mutex added for thenext_update).Also improves (hopefully) the logic when receiving a call-me-maybe: It now clears the
recent_pongfor all addresses not included in the call-me-maybe, and clears thebest_addrtrust if not included in the call-me-maybe.Notes & open questions
Change checklist