From 17f465d5c1b2789d7d3503f18a0f10d37296652f Mon Sep 17 00:00:00 2001 From: SangBin Cho Date: Thu, 27 Aug 2020 12:53:18 -0700 Subject: [PATCH] [Core] Improve raylet failure error msg (#10345) * Improve error message. * Lint. * Addressed code review. --- src/ray/raylet/node_manager.cc | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/ray/raylet/node_manager.cc b/src/ray/raylet/node_manager.cc index f8a445a00..3007da6db 100644 --- a/src/ray/raylet/node_manager.cc +++ b/src/ray/raylet/node_manager.cc @@ -766,7 +766,10 @@ void NodeManager::NodeRemoved(const GcsNodeInfo &node_info) { RAY_CHECK(node_id != self_node_id_) << "Exiting because this node manager has mistakenly been marked dead by the " - << "monitor."; + << "monitor: GCS didn't receive heartbeats within timeout " + << RayConfig::instance().num_heartbeats_timeout() * + RayConfig::instance().raylet_heartbeat_timeout_milliseconds() + << " ms. This is likely since the machine or raylet became overloaded."; // Below, when we remove node_id from all of these data structures, we could // check that it is actually removed, or log a warning otherwise, but that may