Skip to content

WIP Fix too many conn handling#232

Open
Fi3 wants to merge 2 commits into
dmnd-pool:masterfrom
Fi3:FixTooManyConnHandling
Open

WIP Fix too many conn handling#232
Fi3 wants to merge 2 commits into
dmnd-pool:masterfrom
Fi3:FixTooManyConnHandling

Conversation

@Fi3
Copy link
Copy Markdown
Contributor

@Fi3 Fi3 commented Apr 26, 2026

No description provided.

jbesraa and others added 2 commits April 26, 2026 13:56
Drain the pending shares buffer atomically each tick and POST in
batches of 500 (~75-135 KB JSON, well under Rocket's 1 MiB default
JSON limit). Failed chunks are requeued at the head of the buffer
to be retried on the next tick instead of being silently retained
on a single all-or-nothing request that would grow unbounded if the
endpoint ever rejected the payload.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Use a configurable permit pool and accept backoff so new SV1 miners are only
admitted when the proxy still has capacity. Report capacity exhaustion from the
health check and fail fast when the translator handoff queue is saturated
instead of letting ingress block.

Make stats collection awaitable so the overload decision uses the current
downstream count, and move routine connection lifecycle logging to debug so the
real overload signals stay visible. Skip disconnect activity for unnamed
workers to avoid noisy monitor events.
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