[BugFix] add some defensive codes to table metrics

be/src/common/config.h

@@ -1563,4 +1563,6 @@ CONF_mInt32(json_parse_many_batch_size, "1000000");
 CONF_mBool(enable_dynamic_batch_size_for_json_parse_many, "true");
 CONF_mInt32(put_combined_txn_log_thread_pool_num_max, "64");
 CONF_mBool(enable_put_combinded_txn_log_parallel, "false");
+// used to control whether the metrics/ interface collects table metrics
+CONF_mBool(enable_collect_table_metrics, "true");
 } // namespace starrocks::config

be/src/exec/tablet_sink.cpp

@@ -65,7 +65,6 @@
 #include "util/compression/compression_utils.h"
 #include "util/defer_op.h"
 #include "util/stack_util.h"
-#include "util/starrocks_metrics.h"
 #include "util/thread.h"
 #include "util/thrift_rpc_helper.h"
 #include "util/uid_util.h"

be/src/http/action/metrics_action.cpp

@@ -117,6 +117,7 @@ const std::string SimpleCoreMetricsVisitor::MAX_DISK_IO_UTIL_PERCENT = "max_disk
 const std::string SimpleCoreMetricsVisitor::MAX_NETWORK_SEND_BYTES_RATE = "max_network_send_bytes_rate";
 const std::string SimpleCoreMetricsVisitor::MAX_NETWORK_RECEIVE_BYTES_RATE = "max_network_receive_bytes_rate";
 
+const std::string TableMetricsPrefix = "table_";
 void PrometheusMetricsVisitor::visit(const std::string& prefix, const std::string& name, MetricCollector* collector) {
     if (collector->empty() || name.empty()) {
         return;
@@ -127,6 +128,9 @@ void PrometheusMetricsVisitor::visit(const std::string& prefix, const std::strin
     } else {
         metric_name = prefix + "_" + name;
     }
+    if (!config::enable_collect_table_metrics && name.starts_with(TableMetricsPrefix)) {
+        return;
+    }
     // Output metric type
     _ss << "# TYPE " << metric_name << " " << collector->type() << "\n";
     switch (collector->type()) {
@@ -290,6 +294,9 @@ void JsonMetricsVisitor::visit(const std::string& prefix, const std::string& nam
     if (collector->empty() || name.empty()) {
         return;
     }
+    if (!config::enable_collect_table_metrics && name.starts_with(TableMetricsPrefix)) {
+        return;
+    }
 
     rapidjson::Document::AllocatorType& allocator = doc.GetAllocator();
     switch (collector->type()) {

be/src/service/staros_worker.cpp

@@ -81,10 +81,21 @@ StarOSWorker::StarOSWorker() : _mtx(), _shards(), _fs_cache(new_lru_cache(1024))
 
 StarOSWorker::~StarOSWorker() = default;
 
+static const uint64_t kUnknownTableId = UINT64_MAX;
 uint64_t StarOSWorker::get_table_id(const ShardInfo& shard) {
     const auto& properties = shard.properties;
-    CHECK(properties.contains("tableId"));
-    return std::stoull(properties.at("tableId"));
+    auto iter = properties.find("tableId");
+    if (iter == properties.end()) {
+        DCHECK(false) << "tableId doesn't exist in shard properties";
+        return kUnknownTableId;
+    }
+    const auto& tableId = properties.at("tableId");
+    try {
+        return std::stoull(tableId);
+    } catch (const std::exception& e) {
+        DCHECK(false) << "failed to parse tableId: " << tableId << ", " << e.what();
+        return kUnknownTableId;
+    }
 }
 
 absl::Status StarOSWorker::add_shard(const ShardInfo& shard) {

be/src/util/table_metrics.h

@@ -69,8 +69,11 @@ class TableMetricsManager {
 
     TableMetricsPtr get_table_metrics(uint64_t table_id) {
         std::shared_lock l(_mu);
-        DCHECK(_metrics_map.contains(table_id));
-        return _metrics_map.at(table_id);
+        auto iter = _metrics_map.find(table_id);
+        if (iter != _metrics_map.end()) {
+            return iter->second;
+        }
+        return _blackhole_metrics;
     }
 
     void cleanup();
@@ -79,6 +82,11 @@ class TableMetricsManager {
     MetricRegistry* _metrics;
     std::shared_mutex _mu;
     phmap::flat_hash_map<uint64_t, TableMetricsPtr> _metrics_map;
+    // In some cases, we may not be able to obtain the metrics for the corresponding table id,
+    // For example, when drop tablet and data load concurrently,
+    // the Tablets may have been deleted before the load begins, and the table metrics may be cleared.
+    // In such a scenario, we return blackhole metrics to ensure that subsequent processes can work well.
+    TableMetricsPtr _blackhole_metrics = std::make_shared<TableMetrics>();
     // used for cleanup
     int64_t _last_cleanup_ts = 0;