Skip to content

fix(lewis): keep reconnecting to lewis and drop events while disconnected - #182

Draft
ananthb wants to merge 1 commit into
mainfrom
fix/lewis-client-reconnect
Draft

ananthb wants to merge 1 commit into
mainfrom
fix/lewis-client-reconnect

Conversation

@ananthb

@ananthb ananthb commented Oct 9, 2026

Copy link
Copy Markdown
Member
AI description, click for details

The Lewis event client counted every failure over the life of the process and gave up after three, then dropped all events until jet restarted. A Lewis restart, or the keepalive disconnect every ~97 seconds (rpcpool/lewis#65), therefore turned off reporting for good within minutes. While it waited to reconnect, events also queued in the unbounded callback channel.

The client now counts consecutive failed connections and resets the count and the backoff after a connection succeeds. It drops events while it waits between attempts, so memory stays flat during an outage, and max_reconnect_attempts defaults to unlimited. A stream that Lewis ends while jet is running now reconnects; the task returns Ok only when jet's event channel closes, so it can no longer end the task group early.

🤖 Generated with Claude Code

…cted

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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.

1 participant