added initial tracking of namespaces, buckets, keys

1 files changed, 137 insertions, 33 deletions

diff --git a/src/zen/cmds/rpcreplay_cmd.cpp b/src/zen/cmds/rpcreplay_cmd.cpp
index b45f3f50a..231244450 100644
--- a/src/zen/cmds/rpcreplay_cmd.cpp
+++ b/src/zen/cmds/rpcreplay_cmd.cpp
@@ -520,6 +520,130 @@ RpcReplayAnalyzeCommand::Run(const ZenCliOptions& GlobalOptions, int argc, char*
 
 	ZEN_CONSOLE("Recording entry count = {}", EntryCount);
 
+	/////////////////////////////////////////////////////////////////////////
+	//
+	// output schema sketch:
+
+	// namespaces
+	//
+	// <namespace_id>,<namespace>
+
+	struct Namespaces
+	{
+		// NOTE: *NOT* thread safe at this time
+
+		struct NamespaceInfo
+		{
+			int			Id;
+			uint32_t	Hash;
+			std::string Name;
+		};
+
+		std::vector<NamespaceInfo> Namespaces;
+
+		int NamespaceToId(const std::string_view Namespace)
+		{
+			const uint32_t NamespaceHash = HashStringDjb2(Namespace);
+
+			for (const auto& Ns : Namespaces)
+			{
+				if (NamespaceHash == Ns.Hash)
+				{
+					return Ns.Id;
+				}
+			}
+
+			int Id = (int)Namespaces.size();
+
+			Namespaces.push_back(NamespaceInfo{.Id = Id, .Hash = NamespaceHash, .Name = std::string(Namespace)});
+
+			return Id;
+		};
+	};
+
+	Namespaces Ns;
+
+	// buckets
+	//
+	// <bucket_id>,<bucket>
+
+	struct Buckets
+	{
+		// NOTE: *NOT* thread safe at this time
+
+		struct BucketInfo
+		{
+			int			Id;
+			uint32_t	Hash;
+			std::string Name;
+		};
+
+		std::vector<BucketInfo> Buckets;
+
+		int BucketToId(const std::string_view Bucket)
+		{
+			const uint32_t BucketHash = HashStringDjb2(Bucket);
+
+			for (const auto& Ns : Buckets)
+			{
+				if (BucketHash == Ns.Hash)
+				{
+					return Ns.Id;
+				}
+			}
+
+			int Id = (int)Buckets.size();
+
+			Buckets.push_back(BucketInfo{.Id = Id, .Hash = BucketHash, .Name = std::string(Bucket)});
+
+			return Id;
+		};
+	};
+
+	Buckets Bs;
+
+	// keys
+	//
+	// <key_id>,<bucket_id>,<key>
+
+	struct Keys
+	{
+		// NOTE: *NOT* thread safe at this time
+
+		struct KeyInfo
+		{
+			uint32_t Id;
+			uint32_t BucketId;
+			IoHash	 Hash;
+		};
+
+		std::vector<KeyInfo> Keys;
+
+		int KeyToId(uint32_t BucketId, const IoHash& Hash)
+		{
+			ZEN_UNUSED(BucketId, Hash);
+			return 0;
+		};
+	};
+
+	Keys Ks;
+
+	// sessions
+	//
+	// <session_id>,<session>
+
+	// rpc_requests - one entry per recorded RPC request
+	//
+	// <rpc_id>,<session_id>,<namespace_id>,<method>,<request_size>,<attachment_count>
+
+	// payloads - one entry per package attachment, one entry per unstructured value
+	//
+	// <rpc_id>,<payload_index>,<payload_hash>,<payload_size>
+
+	// requests
+	//
+	// <request_id>,<rpc_id>,<key_id>,<payload_hash>
+
 	const uint64_t RangeStart = m_Offset;
 	const uint64_t RangeEnd	  = m_Limit ? (m_Offset + m_Limit) : EntryCount;
 
@@ -581,11 +705,23 @@ RpcReplayAnalyzeCommand::Run(const ZenCliOptions& GlobalOptions, int argc, char*
 						return;
 					}
 
+					const int NsId = Ns.NamespaceToId(Gcv.Namespace);
+
 					const auto BatchSize = Gcv.Requests.size();
 
 					for (const auto& Req : Gcv.Requests)
 					{
-						ZEN_CONSOLE("[{}] GCV {}/{}", BatchSize, Req.Key.Bucket, Req.Key.Hash);
+						const int BsId = Bs.BucketToId(Req.Key.Bucket);
+						const int KId  = Ks.KeyToId(BsId, Req.Key.Hash);
+
+						ZEN_CONSOLE("[{}] GCV ({}){}/({}){}/({}){}",
+									BatchSize,
+									NsId,
+									Gcv.Namespace,
+									BsId,
+									Req.Key.Bucket,
+									KId,
+									Req.Key.Hash);
 					}
 				}
 				break;
@@ -675,38 +811,6 @@ RpcReplayAnalyzeCommand::Run(const ZenCliOptions& GlobalOptions, int argc, char*
 		}
 	});
 
-	/////////////////////////////////////////////////////////////////////////
-	//
-	// output schema sketch:
-
-	// namespaces
-	//
-	// <namespace_id>,<namespace>
-
-	// buckets
-	//
-	// <bucket_id>,<bucket>
-
-	// keys
-	//
-	// <key_id>,<bucket_id>,<key>
-
-	// sessions
-	//
-	// <session_id>,<session>
-
-	// rpc_requests - one entry per recorded RPC request
-	//
-	// <rpc_id>,<session_id>,<namespace_id>,<method>,<request_size>,<attachment_count>
-
-	// payloads - one entry per package attachment, one entry per unstructured value
-	//
-	// <rpc_id>,<payload_index>,<payload_hash>,<payload_size>
-
-	// requests
-	//
-	// <request_id>,<rpc_id>,<key_id>,<payload_hash>
-
 	return 0;
 }