Problem
Sending a restart_service operation with replicated: true and service: http_workers only restarts the workers on the node specified in the request URL. The operation is not propagated to other cluster nodes.
Steps to reproduce
POST https://<any-cluster-node>:9925/
{
"operation": "restart_service",
"service": "http_workers",
"replicated": true
}Expected: HTTP workers restarted on all nodes in the cluster.
Actual: HTTP workers restarted only on the targeted node.
Likely cause
core/bin/restart.ts has replicated handling that calls replication.sendOperationToNode per node, but this path may not be reached for http_workers specifically, or the http_workers restart may short-circuit the process before the replication loop executes.
🤖 Filed by Claude on behalf of Kris.
Problem
Sending a
restart_serviceoperation withreplicated: trueandservice: http_workersonly restarts the workers on the node specified in the request URL. The operation is not propagated to other cluster nodes.Steps to reproduce
Expected: HTTP workers restarted on all nodes in the cluster.
Actual: HTTP workers restarted only on the targeted node.
Likely cause
core/bin/restart.tshasreplicatedhandling that callsreplication.sendOperationToNodeper node, but this path may not be reached forhttp_workersspecifically, or thehttp_workersrestart may short-circuit the process before the replication loop executes.🤖 Filed by Claude on behalf of Kris.