Unify log and error report for easier trouble shooting

[yecol]

Motivations and proposal

• a unified error code, easy to find a concrete reason for failure.
• easy to identify which engine; Within each engine, it can have its own status code.
• easy to find which machine(node), and easy to get the full log.
• no engine should revise/omit the original error info.

Current status

• vineyard/gie are using status codes.
• coordinator would convert to internal error
• enumerate the status code: TBF

Proposal update:

• Append error code, error type and engine to error message.

** deprecated!!!**

• Each engine can maintain its original status code
• Add a unified EngineIdentifier to indicate the status belong to which engine
• Revise the Status implementation to check the error

Example:

• Status Code (assume is analytical engine)

  Enum class StatusCode {
  Ok = 0;
  Invalid = 1;
  Unknown = 255;
}

• Engine Identifier

Enum class EngineIdentifier {
  null = 0;
  analytical = 1;
  interactive = 2;
  learning = 3;
  vineyard = 4;
  graphar = 5;
  gart = 6;
}

• Status Implementation replace the code_ value type from StatusCode to uint32 , which high 16 bits store the engine identifier, the low 16 bits store the status code

// assume this is Status of analytical engine
class Status {
  Status(uint32_t code, std::string message);

  Status InvalidError(std::string message) {
     return Status(((static_cast<uint32_t>(EngineIdentifier::analytical) << 16) | static_cast<uint32_t>(ErrorCode::Invalid)), message);
  }

  // check is invalid error occur in analytical engine.
  bool IsInvalid() {
    return  static_cast<EngineIdentifier>(code_ >> 16) == EngineIdentifier::analytical && static_cast<ErrorCode>(code_ & 0xFF) == ErrorCode::Invalid; 
  }
   
  // check is error occur in vineyard module.
  bool IsVineyardError() {
    return static_cast<EngineIdentifier>(code_ >> 16) == EngineIdentifier::vineyard;
  }

 private:
  uint32_t code_;
  std::string message;
}

Corner cases

[acezen]

Status codes of each engine

Vineyard

Code	Number	Description
OK	0	Not an error; returned on success.
Invalid	1	Invalid data, for example a string that fails parsing.
KeyError	2	Failed key lookups (e.g. column name in a table).
TypeError	3	Type errors (such as mismatching data types).
IOError	4	IO errors (e.g. Failed to open or read from a file).
EndOfFile	5	Reader reaches at the end of file.
NotImplemented	6	An operation or a combination of operation and data types is unimplemented.
AssertionFailed	7	The condition assertion is false.
UserInputError	8	Invalid user input.
ObjectExists	11	The object exists if the user are still trying to creating such object.
ObjectNotExists	12	The object exists if the user are still trying to creating such object.
ObjectSealed	13	The object has already been sealed.
ObjectNotSealed	14	The object hasn't been sealed yet.
ObjectIsBlob	15	Unsupported operations on blob objects.
ObjectTypeError	16	Cast object to mismatched types.
ObjectSpilled	17	The object has already been spilled.
ObjectNotSpilled	18	The object is not spilled yet.
MetaTreeInvalid	21	Metatree related error occurs.
MetaTreeTypeInvalid	22	The "typename" field in metatree is invalid.
MetaTreeTypeNotExists	23	The "typename" field not exists in metatree.
MetaTreeNameInvalid	24	The "id" field in metatree is invalid.
MetaTreeNameNotExists	25	The "id" field not exists in metatree.
MetaTreeLinkInvalid	26	A field in metatree is expected to be a link but it isn't.
MetaTreeSubtreeNotExists	27	Expected subtree doesn't exist in metatree.
VineyardServerNotReady	31	The requested vineyard server is not ready yet.
ArrowError	32	Arrow related error occurs.
ConnectionFailed	33	Client failed to connect to vineyard server.
ConnectionError	34	Client losts connection to vineyard server.
EtcdError	35	The vineyard server meets an etcd related error.
AlreadyStopped	36	The object has already been stopped.
RedisError	37	The vineyard server meets an redis related error.
NotEnoughMemory	41	The vineyard server cannot allocate more memory blocks.
StreamDrained	42	There's no more chunks in the stream.
StreamFailed	43	The stream has failed.
InvalidStreamState	44	Internal invalid state found for stream.
StreamOpened	45	The stream has been opened.
GlobalObjectInvalid	51	Invalid global object structure.
UnknownError	255	Unknown errors.

[acezen]

GraphAr

Code	Number	Description
OK	0	Not an error; returned on success.
KeyError	1	error status for failed key lookups
TypeError	2	error status for type errors
Invalid	3	error status for invalid data
IndexError	4	error status when an index is out of bounds
OutOfMemory	5	error status for out-of-memory conditions
IOError	6	error status when some IO-related operation failed
YamlError	7	error status some yaml parse related operation failed, use to catch mini-yaml raised error
ArrowError	8	error status when some arrow-related operation failed
UnknownError	9	error status for unknown errors

[acezen]

Interactive engine

Currently, GIE defines Error in each layer:

Compiler:

Code	Number	Description
OK	0	Not an error; returned on success.
AntlrParseException	-	antlr parse error
GraphLabelNotFoundException	-	cannot find given label from the schema
GraphPropertyNotFountException	-	cannot find given property from the schema
GraphMisMatchTypeException	-	cannot apply operations on specific type, i.e. get properties from non-node/edge types.
GraphMisMatchOperandsException	-	cannot apply operands to operator

IR :

Code	Number	Description
OK	0	Not an error; returned on success
ParseExprError	1	Parse an expression error
MissingDataError	2	Missing necessary data
CStringError	3	The error while transforming from C-like string, aka char*
UnknownTypeError	4	The provided data type is unknown
InvalidRangeError	5	The provided range is invalid
NegativeIndexError	6	The given index is negative
BuildJobError	7	Build Physical Plan Error
ParsePbError	8	Parse protobuf error
ParentNotFoundError	9	The parent of an operator cannot be found
ColumnNotExistError	10	A column (property) does not exist in the store
TableNotExistError	11	A table (label) does not exist in the store
TagNotExistError	12	A queried tag has not been specified
UnSupported	13	An unsupported error
Others	14	other errors.

Runtime (no number defined yet):

Code	Number	Description
OK	0	Not an error; returned on success
FnGenDecodeOpError	-	Decode pb structure error
FnGenParseError	-	Parse pb structure error
FnGenUnsupportedError	-	An unsupported error in udf gen
FnExecNullGraphError	-	The graph is not registered in engine
FnExecGetTagError	-	A tagged column does not exist in the data
FnExecStoreError	-	An error occurs when query the storage
FnExecExprEvalError	-	A expression evaluation error
FnExecUnExpectedData	-	The data type is unexpected
FnExecAccumError	-	An error in accumulation
FnExecUnSupported	-	An unsupported error in execution
StoreQueryError	-	An error when querying storage
StoreWriteError	-	An error when write a graph
ExprEvalCastError	-	The error while casting from different data types to a general data type Object
ExprEvalMissingContext	-	Missing context for the certain variable
ExprEvalMissingOperands	-	The error of missing required operands in an arithmetic or logical expression
ExprEvalEmptyExpression	-	An error where an empty expression is to be evaluated
ExprEvalUnexpectedDataType	-	Meant to evaluate a variable, but the data type is unexpected (e.g., not a graph-element)
ExprEvalGetNoneFromContext	-	Get None from Context
ClusterInfoMissing	-	An error that cluster info is missing

[acezen]

GART

Code	Number	Description
OK	0	Not an error; returned on success.
KeyError	1	error status for failed key lookups
TypeError	2	error status for type errors
Invalid	3	error status for invalid data
IndexError	4	error status when an index is out of bounds
OutOfMemory	5	error status for out-of-memory conditions
KafkaConnectError	6	error status for failed Kafka connection
CapturerError	7	error status caused by the log capturer
MsgError	8	error status for failed message handling
ParseFileError	9	error status for parsing files
ParseLogeError	10	error status for parsing logs
OperationError	11	error status for operation-related errors
ConfigError	12	error status for incorrect configurations
PremisionError	13	error status caused incorrect premision of database users
VineyardError	14	error status caused by Vineyard
GrinError	15	error status caused by GRIN
UnknownError	16	error status for unknown errors

[acezen]

Groot

Groot use exception to handle errors

[acezen]

Learning engine

Code	Number	Description
OK	0	Not an error; returned on success.

[acezen]

Analytical Engine

Code	Number	Description
Ok	0	Not an error; returned on success.
IOError	1	error code for IO operation failed
ArrowError	2	arrow related error occur
VineyardError	3	vineyard related error occur
UnspecificError	4	unknown error
DistributedError	5	error for distributed job failed, error messages will be gathered
NetworkError	6	network related error
CommandError	7	op command receive failed
DataTypeError	8	Type errors (such as mismatching data types).
IllegalStateError	9	IllegalState error(such as query return a null context wrapper without error message)
InvalidValueError	10	Invalid value, for example a string that fails parsing.
InvalidOperationError	11	Invalid operation error, for example some operation is only available in certain fragment but called by other fragment
UnsupportedOperationError	12	Unsupported operation error
UnimplementedMethod	13	method unimplemented
GraphArError	14	GraphAr related error

[yecol]

• [ ] appends errors and all details (esp. from v6d, in coordinator...)
• [ ] log format in config
• [ ] review boost::leaf