Detect if the background processor exits early and shutdown if so #109
Conversation
valentinewallace
force-pushed
the
2023-07-detect-bgproc-exit
branch
from
662faec
to
bbac222
Compare
src/main.rs
Outdated
| // Exit if CLI polling exits or the background processor exits (which shouldn't happen unless we | ||
| // fail to write to the filesystem). | ||
| tokio::select! { | ||
| _ = cli::poll_for_user_input( |
There was a problem hiding this comment.
Not sure if it is strictly required, but you may want to spawn this instead.
https://docs.rs/tokio/latest/tokio/macro.select.html#runtime-characteristics
There was a problem hiding this comment.
Based on those docs, I can't tell if with the way I had it written previously we would have gotten stuck on the CLI select branch. LMK if you know, for my understanding.
There was a problem hiding this comment.
Probably not stuck given the background processor task was already spawned and would have been running that future already.
There was a problem hiding this comment.
Ah, I meant whether we would ever hit the other select branch to check for the bg processor exiting. Looks like it does work: https://play.rust-lang.org/?version=stable&mode=debug&edition=2021&gist=0c2f28caa7c80e47d969465ba17e9986
valentinewallace
force-pushed
the
2023-07-detect-bgproc-exit
branch
2 times, most recently
from
94dfd54
to
dabe1c3
Compare
src/main.rs
Outdated
| // Exit if CLI polling exits or the background processor exits (which shouldn't happen unless we | ||
| // fail to write to the filesystem). | ||
| tokio::select! { | ||
| _ = cli::poll_for_user_input( |
There was a problem hiding this comment.
Probably not stuck given the background processor task was already spawned and would have been running that future already.
valentinewallace
force-pushed
the
2023-07-detect-bgproc-exit
branch
from
dabe1c3
to
508c202
Compare
| _ = cli_poll => {}, | ||
| bg_res = &mut background_processor => { | ||
| stop_listen_connect.store(true, Ordering::Release); | ||
| peer_manager.disconnect_all_peers(); |
There was a problem hiding this comment.
We should do a last-ditch ChannelManager persist attempt before panicking.
A user ran into an issue where the
BackgroundProcessorsilently stopped, leading toChannelManagerpersistence falling behind + channel force closures. Now we'll detect if the bg processor exits early and log the result.Still need to do a bit more reading up on tokio docs to vet the patch but feedback welcome.