Skip to content

fix failure to periodically flush after dropping a Db instance#1159

Merged
spacejam merged 17 commits intomasterfrom
tyler_subscriber_send_lock
Sep 10, 2020
Merged

fix failure to periodically flush after dropping a Db instance#1159
spacejam merged 17 commits intomasterfrom
tyler_subscriber_send_lock

Conversation

@spacejam
Copy link
Copy Markdown
Owner

this is the ongoing workspace for fixing the bug found in #1152

@soro
Copy link
Copy Markdown
Contributor

soro commented Sep 4, 2020

I have to say that I can't find another obvious deadlock that would lead to an issue in this particular scenario - the interface is such though that it is pretty easy to deadlock if the thread processing the subscription does anything else involving subscriptions, such as try to subscribe to another prefix. Maybe it would be best just to never block on the sending side and just use try_send? That or simply use a concurrent map as the registry so those locks don't have to be taken while sending is going on.

@spacejam spacejam force-pushed the tyler_subscriber_send_lock branch from d201c04 to 0a7061d Compare September 10, 2020 09:17
@spacejam spacejam merged commit 31765d3 into master Sep 10, 2020
@spacejam spacejam deleted the tyler_subscriber_send_lock branch September 10, 2020 09:58
@spacejam spacejam changed the title debug subscription + transaction deadlock fix failure to periodically flush after dropping a Db instance Sep 10, 2020
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