SERVER-71292 Redact all sensitive information in query #8731

7 files changed, 195 insertions, 24 deletions

diff --git a/src/mongo/bson/bsonobj.cpp b/src/mongo/bson/bsonobj.cpp
index c462031bffd..88715dd68de 100644
--- a/src/mongo/bson/bsonobj.cpp
+++ b/src/mongo/bson/bsonobj.cpp
@@ -145,41 +145,56 @@ BSONObj BSONObj::getOwned(const BSONObj& obj) {
     return obj.getOwned();
 }
 
-BSONObj BSONObj::redact(bool onlyEncryptedFields) const {
+BSONObj BSONObj::redact(bool onlyEncryptedFields,
+                        std::function<std::string(const BSONElement&)> fieldNameRedactor) const {
     _validateUnownedSize(objsize());
 
     // Helper to get an "internal function" to be able to do recursion
     struct redactor {
-        void appendRedactedElem(BSONObjBuilder& builder, const BSONElement& e, bool appendMask) {
+        void appendRedactedElem(BSONObjBuilder& builder,
+                                const StringData& fieldNameString,
+                                bool appendMask) {
             if (appendMask) {
-                builder.append(e.fieldNameStringData(), "###"_sd);
+                builder.append(fieldNameString, "###"_sd);
             } else {
-                builder.appendNull(e.fieldNameStringData());
+                builder.appendNull(fieldNameString);
             }
         }
 
         void operator()(BSONObjBuilder& builder,
                         const BSONObj& obj,
                         bool appendMask,
-                        bool onlyEncryptedFields) {
+                        bool onlyEncryptedFields,
+                        std::function<std::string(const BSONElement&)> fieldNameRedactor) {
             for (BSONElement e : obj) {
+                StringData fieldNameString;
+                // Temporarily allocated string that must live long enough to be copied by builder.
+                std::string tempString;
+                if (!fieldNameRedactor) {
+                    fieldNameString = e.fieldNameStringData();
+                } else {
+                    tempString = fieldNameRedactor(e);
+                    fieldNameString = {tempString};
+                }
                 if (e.type() == Object) {
-                    BSONObjBuilder subBuilder = builder.subobjStart(e.fieldNameStringData());
-                    operator()(subBuilder, e.Obj(), appendMask, onlyEncryptedFields);
+                    BSONObjBuilder subBuilder = builder.subobjStart(fieldNameString);
+                    operator()(
+                        subBuilder, e.Obj(), appendMask, onlyEncryptedFields, fieldNameRedactor);
                     subBuilder.done();
                 } else if (e.type() == Array) {
-                    BSONObjBuilder subBuilder = builder.subarrayStart(e.fieldNameStringData());
-                    operator()(subBuilder, e.Obj(), appendMask, onlyEncryptedFields);
+                    BSONObjBuilder subBuilder = builder.subarrayStart(fieldNameString);
+                    operator()(
+                        subBuilder, e.Obj(), appendMask, onlyEncryptedFields, fieldNameRedactor);
                     subBuilder.done();
                 } else {
                     if (onlyEncryptedFields) {
                         if (e.type() == BinData && e.binDataType() == BinDataType::Encrypt) {
-                            appendRedactedElem(builder, e, appendMask);
+                            appendRedactedElem(builder, fieldNameString, appendMask);
                         } else {
                             builder.append(e);
                         }
                     } else {
-                        appendRedactedElem(builder, e, appendMask);
+                        appendRedactedElem(builder, fieldNameString, appendMask);
                     }
                 }
             }
@@ -188,7 +203,7 @@ BSONObj BSONObj::redact(bool onlyEncryptedFields) const {
 
     try {
         BSONObjBuilder builder;
-        redactor()(builder, *this, /*appendMask=*/true, onlyEncryptedFields);
+        redactor()(builder, *this, /*appendMask=*/true, onlyEncryptedFields, fieldNameRedactor);
         return builder.obj();
     } catch (const ExceptionFor<ErrorCodes::BSONObjectTooLarge>&) {
     }
@@ -198,7 +213,7 @@ BSONObj BSONObj::redact(bool onlyEncryptedFields) const {
     // we use BSONType::jstNull, which ensures the redacted object will not be larger than the
     // original.
     BSONObjBuilder builder;
-    redactor()(builder, *this, /*appendMask=*/false, onlyEncryptedFields);
+    redactor()(builder, *this, /*appendMask=*/false, onlyEncryptedFields, fieldNameRedactor);
     return builder.obj();
 }
 
diff --git a/src/mongo/bson/bsonobj.h b/src/mongo/bson/bsonobj.h
index a0088c8b70d..83453bcc9e1 100644
--- a/src/mongo/bson/bsonobj.h
+++ b/src/mongo/bson/bsonobj.h
@@ -273,7 +273,9 @@ public:
     /**
      * @return a new full (and owned) redacted copy of the object.
      */
-    BSONObj redact(bool onlyEncryptedFields = false) const;
+    BSONObj redact(
+        bool onlyEncryptedFields = false,
+        std::function<std::string(const BSONElement&)> fieldNameRedactor = nullptr) const;
 
     /**
      * Readable representation of a BSON object in an extended JSON-style notation.
diff --git a/src/mongo/db/pipeline/document_source_telemetry.cpp b/src/mongo/db/pipeline/document_source_telemetry.cpp
index 36acd596241..812261edc1d 100644
--- a/src/mongo/db/pipeline/document_source_telemetry.cpp
+++ b/src/mongo/db/pipeline/document_source_telemetry.cpp
@@ -101,8 +101,9 @@ void DocumentSourceTelemetry::buildTelemetryStoreIterator() {
                 const auto partitionReadTime =
                     Timestamp{Timestamp(Date_t::now().toMillisSinceEpoch() / 1000, 0)};
                 for (auto&& [key, metrics] : *partition) {
-                    Document d{
-                        {{"key", key}, {"metrics", metrics.toBSON()}, {"asOf", partitionReadTime}}};
+                    Document d{{{"key", metrics.redactKey(key)},
+                                {"metrics", metrics.toBSON()},
+                                {"asOf", partitionReadTime}}};
                     _queue.push(std::move(d));
                 }
             });
diff --git a/src/mongo/db/query/query_knobs.idl b/src/mongo/db/query/query_knobs.idl
index a4cdc297567..21e86666c2f 100644
--- a/src/mongo/db/query/query_knobs.idl
+++ b/src/mongo/db/query/query_knobs.idl
@@ -51,6 +51,16 @@ enums:
             # Force the Bonsai optimizer for all queries.
             kForceBonsai: "forceBonsai"
 
+    QueryTelemetryFieldNameRedactionStrategy:
+        description: "Enum for possible values of queryFieldNameRedactionStrategy."
+        type: string
+        values:
+            kNoRedactionStrategy: "none"
+            # Use the constant redaction strategy.
+            kConstantRedactionStrategy: "constant"
+            # Use a prefix of sha256 redaction strategy
+            kSha256RedactionStrategy: "sha256"
+
 server_parameters:
 
   #
@@ -973,6 +983,17 @@ server_parameters:
     validator:
       callback: telemetry_util::validateTelemetryStoreSize
 
+  internalQueryConfigureTelemetryFieldNameRedactionStrategy:
+    description:  Choose between the "none" implementation which will not redact any fields names, the "constant"
+      implementation, which will replace all field names with a constant, and the "sha256" implementation, which
+      preserves the identify of a field.
+    set_at: [ startup, runtime ]
+    cpp_class:
+      name: QueryTelemetryControl
+      data: synchronized_value<QueryTelemetryFieldNameRedactionStrategyEnum>
+    default:
+      expr: QueryTelemetryFieldNameRedactionStrategyEnum::kConstantRedactionStrategy
+
 
 # Note for adding additional query knobs:
 #
diff --git a/src/mongo/db/query/telemetry.cpp b/src/mongo/db/query/telemetry.cpp
index 7af325183f1..7b8c2c36c89 100644
--- a/src/mongo/db/query/telemetry.cpp
+++ b/src/mongo/db/query/telemetry.cpp
@@ -184,7 +184,6 @@ bool shouldCollect(const ServiceContext* serviceCtx) {
     if (!telemetryRateLimiter(serviceCtx)->handleRequestSlidingWindow()) {
         return false;
     }
-    // TODO SERVER-71244: check if it's a FLE collection here (maybe pass in the request)
     return true;
 }
 
@@ -195,8 +194,10 @@ void addToFindKey(BSONObjBuilder& builder, const StringData& fieldName, const BS
     serializeBSONWhenNotEmpty(value.redact(false), fieldName, &builder);
 }
 
-// Call this function from inside the redact() function on every BSONElement in the BSONObj.
-void throwIfEncounteringFLEPayload(BSONElement e) {
+/**
+ * Recognize FLE payloads in a query and throw an exception if found.
+ */
+void throwIfEncounteringFLEPayload(const BSONElement& e) {
     constexpr auto safeContentLabel = "__safeContent__"_sd;
     constexpr auto fieldpath = "$__safeContent__"_sd;
     if (e.type() == BSONType::Object) {
@@ -204,12 +205,12 @@ void throwIfEncounteringFLEPayload(BSONElement e) {
         uassert(ErrorCodes::EncounteredFLEPayloadWhileRedacting,
                 "Encountered __safeContent__, or an $_internalFle operator, which indicate a "
                 "rewritten FLE2 query.",
-                fieldname == safeContentLabel || fieldname.startsWith("$_internalFle"_sd));
+                fieldname != safeContentLabel && !fieldname.startsWith("$_internalFle"_sd));
     } else if (e.type() == BSONType::String) {
         auto val = e.valueStringData();
         uassert(ErrorCodes::EncounteredFLEPayloadWhileRedacting,
                 "Encountered $__safeContent__ fieldpath, which indicates a rewritten FLE2 query.",
-                val == fieldpath);
+                val != fieldpath);
     } else if (e.type() == BSONType::BinData && e.isBinData(BinDataType::Encrypt)) {
         int len;
         auto data = e.binData(len);
@@ -263,8 +264,104 @@ private:
     TelemetryStore::Partition _partitionLock;
 };
 
+/**
+ * Upon reading telemetry data, we redact some keys. This is the list. See
+ * TelemetryMetrics::redactKey().
+ */
+const stdx::unordered_set<std::string> kKeysToRedact = {"pipeline", "find"};
+
+std::string sha256FieldNameHasher(const BSONElement& e) {
+    auto&& fieldName = e.fieldNameStringData();
+    auto hash = SHA256Block::computeHash({ConstDataRange(fieldName.rawData(), fieldName.size())});
+    return hash.toString().substr(0, 12);
+}
+
+std::string constantFieldNameHasher(const BSONElement& e) {
+    return {"###"};
+}
+
+/**
+ * Admittedly an abuse of the BSON redaction interface, we recognize FLE payloads here and avoid
+ * collecting telemetry for the query.
+ */
+std::string fleSafeFieldNameRedactor(const BSONElement& e) {
+    throwIfEncounteringFLEPayload(e);
+    // Ideally we would change interface to avoid copying here.
+    return e.fieldNameStringData().toString();
+}
+
 }  // namespace
 
+const BSONObj& TelemetryMetrics::redactKey(const BSONObj& key) const {
+    if (_redactedKey) {
+        return *_redactedKey;
+    }
+
+    auto redactionStrategy = ServerParameterSet::getNodeParameterSet()
+                                 ->get<QueryTelemetryControl>(
+                                     "internalQueryConfigureTelemetryFieldNameRedactionStrategy")
+                                 ->_data.get();
+
+    // The telemetry key is of the following form:
+    // { "<CMD_TYPE>": {...}, "namespace": "...", "applicationName": "...", ... }
+    //
+    // The part of the key we need to redact is the object in the <CMD_TYPE> element. In the case of
+    // an aggregate() command, it will look something like:
+    // > "pipeline" : [ { "$telemetry" : {} },
+    //					{ "$addFields" : { "x" : { "$someExpr" {} } } } ],
+    // We should preserve the top-level stage names in the pipeline but redact all field names of
+    // children.
+    //
+    // The find-specific key will look like so:
+    // > "find" : { "find" : "###", "filter" : { "_id" : { "$ne" : "###" } } },
+    // Again, we should preserve the top-level keys and redact all field names of children.
+    BSONObjBuilder redacted;
+    for (BSONElement e : key) {
+        if ((e.type() == Object || e.type() == Array) &&
+            kKeysToRedact.count(e.fieldNameStringData().toString()) == 1) {
+            auto redactor = [&](BSONObjBuilder subObj, const BSONObj& obj) {
+                for (BSONElement e2 : obj) {
+                    if (e2.type() == Object) {
+                        switch (redactionStrategy) {
+                            case QueryTelemetryFieldNameRedactionStrategyEnum::
+                                kSha256RedactionStrategy:
+                                subObj.append(e2.fieldNameStringData(),
+                                              e2.Obj().redact(false, sha256FieldNameHasher));
+                                break;
+                            case QueryTelemetryFieldNameRedactionStrategyEnum::
+                                kConstantRedactionStrategy:
+                                subObj.append(e2.fieldNameStringData(),
+                                              e2.Obj().redact(false, constantFieldNameHasher));
+                                break;
+                            case QueryTelemetryFieldNameRedactionStrategyEnum::kNoRedactionStrategy:
+                                subObj.append(e2.fieldNameStringData(), e2.Obj().redact(false));
+                                break;
+                        }
+                    } else {
+                        subObj.append(e2);
+                    }
+                }
+                subObj.done();
+            };
+
+            // Now we're inside the <CMD_TYPE>:{} entry and want to preserve the top-level field
+            // names. If it's a [pipeline] array, we redact each element in isolation.
+            if (e.type() == Object) {
+                redactor(redacted.subobjStart(e.fieldNameStringData()), e.Obj());
+            } else {
+                BSONObjBuilder subArr = redacted.subarrayStart(e.fieldNameStringData());
+                for (BSONElement stage : e.Obj()) {
+                    redactor(subArr.subobjStart(""), stage.Obj());
+                }
+            }
+        } else {
+            redacted.append(e);
+        }
+    }
+    _redactedKey = redacted.obj();
+    return *_redactedKey;
+}
+
 void registerAggRequest(const AggregateCommandRequest& request, OperationContext* opCtx) {
     if (request.getEncryptionInformation()) {
         return;
@@ -285,7 +382,8 @@ void registerAggRequest(const AggregateCommandRequest& request, OperationContext
         for (auto&& stage : request.getPipeline()) {
             auto el = stage.firstElement();
             BSONObjBuilder stageBuilder = pipelineBuilder.subobjStart("stage"_sd);
-            stageBuilder.append(el.fieldNameStringData(), el.Obj().redact(false));
+            stageBuilder.append(el.fieldNameStringData(),
+                                el.Obj().redact(false, fleSafeFieldNameRedactor));
             stageBuilder.done();
         }
         pipelineBuilder.done();
@@ -340,7 +438,8 @@ void registerFindRequest(const FindCommandRequest& request,
         auto findBson = request.toBSON({});
         for (auto&& findEntry : findBson) {
             if (findEntry.isABSONObj()) {
-                telemetryKey.append(findEntry.fieldNameStringData(), findEntry.Obj().redact(false));
+                telemetryKey.append(findEntry.fieldNameStringData(),
+                                    findEntry.Obj().redact(false, fleSafeFieldNameRedactor));
             } else {
                 telemetryKey.append(findEntry.fieldNameStringData(), "###"_sd);
             }
diff --git a/src/mongo/db/query/telemetry.h b/src/mongo/db/query/telemetry.h
index 3a2f41a7e4e..0e225d2f17e 100644
--- a/src/mongo/db/query/telemetry.h
+++ b/src/mongo/db/query/telemetry.h
@@ -107,6 +107,11 @@ public:
     }
 
     /**
+     * Redact a given telemetry key.
+     */
+    const BSONObj& redactKey(const BSONObj& key) const;
+
+    /**
      * Timestamp for when this query shape was added to the store. Set on construction.
      */
     const Timestamp firstSeenTimestamp;
@@ -130,6 +135,12 @@ public:
     AggregatedMetric docsScanned;
 
     AggregatedMetric keysScanned;
+
+private:
+    /**
+     * We cache the redacted key the first time it's computed.
+     */
+    mutable boost::optional<BSONObj> _redactedKey;
 };
 
 struct TelemetryPartitioner {
@@ -139,9 +150,15 @@ struct TelemetryPartitioner {
     }
 };
 
+/**
+ * Average key size used to pad the metrics size. We store a cached redaction of the key in the
+ * TelemetryMetrics object.
+ */
+const size_t kAverageKeySize = 100;
+
 struct ComputeEntrySize {
     size_t operator()(const TelemetryMetrics& entry) {
-        return sizeof(TelemetryMetrics);
+        return sizeof(TelemetryMetrics) + kAverageKeySize;
     }
 };
 
diff --git a/src/mongo/db/query/telemetry_util.cpp b/src/mongo/db/query/telemetry_util.cpp
index dcce8d7452b..fea9ad6587b 100644
--- a/src/mongo/db/query/telemetry_util.cpp
+++ b/src/mongo/db/query/telemetry_util.cpp
@@ -32,6 +32,7 @@
 #include "mongo/base/status.h"
 #include "mongo/db/concurrency/d_concurrency.h"
 #include "mongo/db/query/partitioned_cache.h"
+#include "mongo/db/query/query_knobs_gen.h"
 #include "mongo/db/query/util/memory_util.h"
 #include "mongo/db/service_context.h"
 #include "mongo/logv2/log.h"
@@ -82,3 +83,18 @@ const Decorable<ServiceContext>::Decoration<std::unique_ptr<OnParamChangeUpdater
     telemetryStoreOnParamChangeUpdater =
         ServiceContext::declareDecoration<std::unique_ptr<OnParamChangeUpdater>>();
 }  // namespace mongo::telemetry_util
+
+namespace mongo {
+void QueryTelemetryControl::append(OperationContext*,
+                                   BSONObjBuilder* b,
+                                   StringData name,
+                                   const boost::optional<TenantId>&) {
+    *b << name << QueryTelemetryFieldNameRedactionStrategy_serializer(_data.get());
+}
+
+Status QueryTelemetryControl::setFromString(StringData value, const boost::optional<TenantId>&) {
+    _data = QueryTelemetryFieldNameRedactionStrategy_parse(
+        IDLParserContext("internalQueryConfigureTelemetryFieldNameRedactionStrategy"), value);
+    return Status::OK();
+}
+}  // namespace mongo