Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

mixclient: Remove submit queue channel #3401

Merged
merged 2 commits into from
Jul 6, 2024
Merged

Conversation

jrick
Copy link
Member

@jrick jrick commented Jul 5, 2024

The submit queue channel was not actually increasing any performance. (*peer).submit() would synchronously wait for all error results, and the call to (*Wallet).SubmitMixMessage was already synchronized by the mixpool mutex.

Furthermore, this also fixes a deadlock that was observed after a mixing wallet with the RPC syncer mode reconnected to a restarted dcrd. Pair request messages were being submitted onto the channel with the client mutex held in (*Client).Dicemix. However, handleSubmitQueue had already exited and the client had not yet been restarted after dcrd reconnect, and was unable to be started due to the locked mutex.

The submit queue channel was not actually increasing any performance.
(*peer).submit() would synchronously wait for all error results, and the call
to (*Wallet).SubmitMixMessage was already synchronized by the mixpool mutex.

Furthermore, this also fixes a deadlock that was observed after a mixing
wallet with the RPC syncer mode reconnected to a restarted dcrd.  Pair request
messages were being submitted onto the channel with the client mutex held in
(*Client).Dicemix.  However, handleSubmitQueue had already exited and the
client had not yet been restarted after dcrd reconnect, and was unable to be
started due to the locked mutex.
The client mutex was being held during the initial publishing of pair request
messages to prevent a situation where a submit errored but the peer was still
associated with the client, or the message was submitted to mixpool and sent
to other peers but our local peer had not yet been associated to the client at
the time an epoch tick occurred.  This should not be a situation we will
encounter anymore, since the PR submissions are spaced out intentionally to
avoid sending them close to the epoch.
@davecgh davecgh added this to the 2.1.0 milestone Jul 5, 2024
@davecgh davecgh merged commit b5c58ed into decred:master Jul 6, 2024
2 checks passed
@jrick jrick deleted the nosubmitqueue branch July 6, 2024 03:02
@davecgh davecgh modified the milestones: 2.1.0, 2.0.4 Aug 28, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants