Conversation
Signed-off-by: Drew Newberry <anewberry@nvidia.com>
|
Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually. Contributors can view more details about this message here. |
|
Cross-reference: #4011 has been approved and submitted to the merge queue. It covers the pending-relay reconnect recovery originally implemented in #3583 and restacked here, together with duplicate-delivery suppression from #3588/#3619. It adds atomic enqueue/session validation, a shared setup deadline, SSH proxy shutdown/error propagation, and bounded Docker diagnostics. Draft #3766 remains the follow-up for deterministic reconnect conformance coverage: rebase it after #4011 merges and remove the overlapping server fix while retaining the tests and architecture documentation. The VM fixture changes here are separate from #4011, and the underlying Docker startup stall remains unresolved. |
Summary
Recover relay opens after supervisor reconnect and prevent VM proxy fixture port collisions. Clean restack of #3583.
Related Issue
No issue required: localized fixes for the Docker sync and VM proxy failures cited in #3583.
Changes
Testing
mise run pre-commitpassed on this restacked branch.mise run cifails in Go gateway-list tests because this host has /etc/openshell/gateways/default; see test(go): isolate gateway listing from host config #3590.Checklist