From 1d225a1c6c16157282e1937cfb054ee4971c3460 Mon Sep 17 00:00:00 2001
From: Andrei Rykov <arykov@ydb.tech>
Date: Wed, 17 Jan 2024 06:27:05 +0000
Subject: [PATCH 1/2] added RestartsPerPeriod info

---
 ydb/core/health_check/health_check.cpp | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/ydb/core/health_check/health_check.cpp b/ydb/core/health_check/health_check.cpp
index e4bce7b539d8..1f64437bbd2c 100644
--- a/ydb/core/health_check/health_check.cpp
+++ b/ydb/core/health_check/health_check.cpp
@@ -230,6 +230,7 @@ class TSelfCheckRequest : public TActorBootstrapped<TSelfCheckRequest> {
         TVector<TString> StoragePoolNames;
         THashMap<std::pair<TTabletId, NNodeWhiteboard::TFollowerId>, const NKikimrHive::TTabletInfo*> MergedTabletState;
         THashMap<TNodeId, TNodeTabletState> MergedNodeTabletState;
+        THashMap<TNodeId, ui32> RestartsPerPeriod;
         ui64 StorageQuota;
         ui64 StorageUsage;
     };
@@ -1056,6 +1057,7 @@ class TSelfCheckRequest : public TActorBootstrapped<TSelfCheckRequest> {
                             TString path(itFilterDomainKey->second);
                             TDatabaseState& state(DatabaseState[path]);
                             state.ComputeNodeIds.emplace_back(hiveStat.GetNodeId());
+                            state.RestartsPerPeriod[hiveStat.GetNodeId()] = hiveStat.GetRestartsPerPeriod();
                         }
                     }
                 }

From 3cd00afee696b3c778c8c3a72a02c30d1c456eb2 Mon Sep 17 00:00:00 2001
From: Andrei Rykov <arykov@ydb.tech>
Date: Wed, 17 Jan 2024 06:57:18 +0000
Subject: [PATCH 2/2] added issue

---
 ydb/core/health_check/health_check.cpp | 9 +++++++--
 1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/ydb/core/health_check/health_check.cpp b/ydb/core/health_check/health_check.cpp
index 1f64437bbd2c..4f2fcc5258cf 100644
--- a/ydb/core/health_check/health_check.cpp
+++ b/ydb/core/health_check/health_check.cpp
@@ -1248,9 +1248,13 @@ class TSelfCheckRequest : public TActorBootstrapped<TSelfCheckRequest> {
         }
     }
 
-    void FillComputeNodeStatus(TNodeId nodeId, Ydb::Monitoring::ComputeNodeStatus& computeNodeStatus, TSelfCheckContext context) {
+    void FillComputeNodeStatus(TDatabaseState& databaseState,TNodeId nodeId, Ydb::Monitoring::ComputeNodeStatus& computeNodeStatus, TSelfCheckContext context) {
         FillNodeInfo(nodeId, context.Location.mutable_compute()->mutable_node());
 
+        if (databaseState.RestartsPerPeriod[nodeId] > 30) {
+            context.ReportStatus(Ydb::Monitoring::StatusFlag::RED, "Node is restarting too often", ETags::NodeState);
+        }
+
         auto itNodeSystemState = MergedNodeSystemState.find(nodeId);
         if (itNodeSystemState != MergedNodeSystemState.end()) {
             const NKikimrWhiteboard::TSystemStateInfo& nodeSystemState(*itNodeSystemState->second);
@@ -1308,8 +1312,9 @@ class TSelfCheckRequest : public TActorBootstrapped<TSelfCheckRequest> {
             }
             for (TNodeId nodeId : *computeNodeIds) {
                 auto& computeNode = *computeStatus.add_nodes();
-                FillComputeNodeStatus(nodeId, computeNode, {&context, "COMPUTE_NODE"});
+                FillComputeNodeStatus(databaseState, nodeId, computeNode, {&context, "COMPUTE_NODE"});
             }
+            context.ReportWithMaxChildStatus("Some nodes are restarting too often", ETags::ComputeState, {ETags::NodeState});
             context.ReportWithMaxChildStatus("Compute is overloaded", ETags::ComputeState, {ETags::OverloadState});
             Ydb::Monitoring::StatusFlag::Status tabletsStatus = Ydb::Monitoring::StatusFlag::GREEN;
             computeNodeIds->push_back(0); // for tablets without node