Default package

Get desktop application:
View/edit binary Protocol Buffers messages

Alpha services are deprecated. Please use SearchV1

• rpc DeleteCustomSchema (CustomIndexSchemaDel, google.protobuf.Empty)

  router.proto:679

  for custom schema deletion This is alpha version of this api and doesn't synchronize across apiserver instances. apiserver restart is needed to get updated schema information from hbase.

• rpc Search (SearchRequest, stream NotificationBatch)

  router.proto:664

  you know, for search...

• rpc SearchSubscribe (SearchRequest, stream NotificationBatch)

  router.proto:667

  SearchSubscribe allows the client to request a live stream of updates based on client search request

• rpc SearchWithAggregation (SearchRequestWithAggr, AggrResponse)

  router.proto:669

  for search with aggregation

• rpc SearchWithAggregationStream (SearchRequestWithAggr, stream ByteStream)

  router.proto:673

  SearchWithAggregationStream sends the protobuf-serialized form of AggrResponse as in a stream of byteArrays. Receiver needs to append the "bytearrays", and protobuf-deserialize to obtain the result. Intended for messages exceeding the grpc size limit

• rpc SetCustomSchema (CustomIndexSchema, google.protobuf.Empty)

  router.proto:675

  for custom schema configuration

• rpc CreateDataset (CreateDatasetRequest, google.protobuf.Empty)

  router.proto:624

  CreateDataset from a given Dataset wrapped in a CreateDatasetRequest

  message CreateDatasetRequest

  router.proto:453

  • optional Dataset dataset = 1

• rpc CreateSession (CreateSessionRequest, stream CreateSessionResponse)

  router.proto:637

  CreateSession creates session for user

  message CreateSessionRequest

  router.proto:508

  CreateSessionRequest is used to create session for user

  • optional Dataset dataset = 1

  • int64 timeout = 2

  message CreateSessionResponse

  router.proto:514

  CreateSessionResponse returns the issued JWT token

  • string jwtToken = 1

  • int64 expiry = 2

• rpc GetPermissionSet (GetRequest, stream PermissionSet)

  router.proto:631

  GetPermissionSet returns the set of all permissions present for the datasets specified in the 'query'(s) of the GetRequest.

  message PermissionSet

  router.proto:490

  A PermissionSet can be represented as a list of SetPermissionRequests

  • repeated SetPermissionRequest permissions = 1

• rpc SetPassword (SetPasswordRequest, google.protobuf.Empty)

  router.proto:634

  SetPassword sets the password for a user.

  message SetPasswordRequest

  router.proto:502

  SetPasswordRequest is used to set the password for a dataset.

  • optional Dataset dataset = 1

  • string password = 2

• rpc SetPermission (SetPermissionRequest, google.protobuf.Empty)

  router.proto:627

  SetPermission sets a permission for a dataset using a SetPermissionRequest.

• rpc Enroll (CSR, CRT)

  ca.proto:21

  Enroll is called by client passing Certificate Signing Request (CSR) containing its personal certificate to be signed. The server returns back the certificate (CRT) to authenticate into IngestGateway or Router.

• rpc Reenroll (CSR, CRT)

  ca.proto:26

  Reenroll is called by client passing Certificate Signing Request (CSR) when it wants to renew its personal certificate. The server returns a new certificate (CRT).

Cluster service gives some descriptions about the cluster where the service is running.

• rpc ClusterInfo (google.protobuf.Empty, stream ClusterDescription)

  router.proto:689

  message ClusterDescription

  router.proto:494

  • optional google.protobuf.Timestamp timestamp = 1

    timestamp represent the server's current time.

  • string clusterName = 2

  • uint32 epoch = 3

• rpc SQL (SQLRequest, stream SQLResponse)

  router.proto:683

  message SQLRequest

  router.proto:519

  • string query = 1

  • repeated bytes args = 2

    Neat encoded values for the arguments to the stmt query

  message SQLResponse

  router.proto:530

  • repeated SQLResponseRow rows = 1

  • optional SQLResponse.Metadata metadata = 2

• rpc Get (GetRequest, stream NotificationBatch)

  router.proto:606

  Get is used to request notifications for a given path over a specified time range. Wildcards are supported with Get requests, but when given a range of time the server will resolve all wildcard paths at the starting timestamp of the given range, so any pointers and/or paths that are created after the given start timestamp will not be accounted for during wildcard resolution. The client may receive duplicate notifications.

• rpc GetAndSubscribe (GetAndSubscribeRequest, stream NotificationBatch)

  router.proto:617

  GetAndSubscribe allows the client to issue one request to do both Get and Subscribe requests. The server will first send a mix of subscribe and get batches, and there's no distinction between which batches are subscribe or get batches. Then the server will send a sync signal signaling that the Get stream has finished. After that, server will stream out only subscribe batches. There's no order guarantee for batches received by client. The end of get stream sync signal will be in the Metadata field in the NotificationBatch, as 1 key, value pair: [GetRequest:EOF]. If there are batches returned from Get request, this metadata will be in the last Get batch; if there's no results from Get, it will be sent in an empty NotificationBatch with only the Metadata field.

  message GetAndSubscribeRequest

  router.proto:126

  • repeated Query query = 1

  • uint64 start = 2

    (Optional) Lower bound (inclusive) of time range to retrieve, in nanoseconds since the Epoch. If `start` is set, the returned results will correspond to a time range query (from `start` to `end` or now, depending on whether `end` is set or not). If `start` is not set, this request will execute a "limit" query based on `versions`. The value 0 corresponds to an unset field.

  • uint32 versions = 3

    (Optional) Maximum number of versions to retrieve. If `start` is set, this value is ignored. Otherwise, will return at most (`versions` + 1) past versions.

  • bool exact_range = 4

    (Optional) Whether to return the initial state of the requested paths at time `start`. The "initial state" is the set of notifications that fully specify the state of the path at that time. It includes all relevant notifications before (and up to) `start`. If `start` is not set, this is ignored.

  • optional Sharding sharded_sub = 5

    (Optional) Number of client instances, id of the client making the request, and sharding method for the implementation of horizontal scaling. If not set, no filtering will happen.

  • PathWildCardExpandType wildcard_expand_type = 6

    (Optional) Specifies wildcard expansion strategy when time range is specified. This option has no effect when time range (start/end) is not used.

• rpc GetDatasets (DatasetsRequest, stream DatasetsResponse)

  router.proto:619

  message DatasetsRequest

  router.proto:443

  DatasetsRequest

  • repeated string types = 1

  • optional Sharding sharded_sub = 2

  • optional Dataset parent = 3

  message DatasetsResponse

  router.proto:449

  • repeated Dataset datasets = 1

• rpc Publish (PublishRequest, google.protobuf.Empty)

  router.proto:588

  Publish is used to send notifications to Cloudvision. They will be saved into the storage and sent to all the clients subscribing to the same device/path. * Publish guarantees atomicity of the data saved per {timestamp+path+key}. For Notification => For one Notification having multiple keys, each key is ensured to be saved atomically but atomicity is not guaranteed for the entire notification. For NotificationBatch => if Notif[1] and Notif[5] both have updates for a {timestamp+path+key} either the update of Notif[1] will be saved, or the update of Notif[5] will be saved. The value will be one or the other, not a corrupted combination of both requests. * There is no guarantee for write order within a single publish request. When sending multiple notifications where multiple notification will have the same timestamp, path and keys, Publish does not guarantee that Notif[1] will be processed before Notif[5] This means that for two notifications in the same Publish call having the same {timestamp+path+key}, the result is undefined and will randomly vary (i.e. the first notif data will be saved, or the second one). The client must send two synchronous Publish requests to guarantee the write order at which the requests are processed. * Publish is asynchronous by default: When the call to Publish ends without error, it means the data has been correctly received by Cloudvision but not stored yet. So, if a "get" call is done right after the Publish call, the get might not return the data just published. When the "sync" field is set to true in PublishRequest, the Publish will be synchronous: When the call to Publish ends without error, it means the data has been correctly received AND stored by Cloudvision. So, if a "get" call is done right after the synchronous Publish call, the get will return the data just published (unless someone else stored more recent data of course). * Client-side and Server-side timestamping: The notification object has a timestamp that can be populated by the client. In case the Client sends a notification with a "null" timestamp as the Notification.timestamp field, the server will populate the timestamp with the current time of the node with the server process is running. This "current time" will be queried once at the beginning of the Publish request and will be used as the Notification.timestamp for all the notification having this field as null.

  message PublishRequest

  router.proto:425

  PublishRequest

  • optional NotificationBatch batch = 1

    The batch of notification sent to Cloudvision

  • bool sync = 2

    Used to have a synchronous or asynchronous write to the Cloudvision storage

  • optional Notification.Update compare = 3

    If provided, the batch notification will be stored only if one of the following conditions are true: a) compare.value matches the current value stored for field compare.key b) compare.value is not provided and the compare.key field does not exist in the specified dataset path The comparison and storage operations are guaranteed to occur atomically per notification, so only one notification is expected in the batch. This parameter can be encoded using EncodeNotificationUpdate or EncodeNotificationUpdateKey for conditions a) and b), respectively.

• rpc Subscribe (SubscribeRequest, stream NotificationBatch)

  router.proto:599

  Subscribe allows the client to request a live stream of updates (V1: either based on regexp or exact match, V2: based on exact match) There is no order guarantee for batches received by subscribers. It means that two batches A and B published synchronously (B is published after A) the subscribers can receive batch A first or B second, OR batch B first and A second. This is also true for notifications within a batch. The backend can decide to split a batch and reorder notifications so subscribers might receive notifications within a batch in a different order that they were published.

  message SubscribeRequest

  router.proto:69

  SubscribeRequest defines the stream subset the client is subscribing to. V1: The filtering is done first by dataset: - if Query.Dataset.Name == "<name>" and Query.Dataset.Typ == "<type>", matching notificaitons from dataset "<type>:<name>" that are returned. - if Query.Dataset.Name == "" and Query.Dataset.Typ == "<type>", matching notifications from datasets with type "<type>" are returned. - if Query.Dataset.Name == "" and Query.Dataset.Typ == "", matching notifications from all datasets are returned. TODO: for now the above rules only work if Path.regexp_path is specified. Path.exact_path expects Query.Dataset.Name and Query.Dataset.Typ to be non-empty If a notification matches any of the rules above, next it's matched against Query.Paths. The shardedsub field contains the information relative to horizontal scaling (number of instances, ID of the subscribing instance, and dispatching method) V2: If path elements are not provided: Uses V1 logic. If path elements are provided: Notification filtering is done by queries. For each query, the dataset name and type must be non-empty. Notifications that match the query paths are streamed to the client. The sharded sub field contains information relating to horizontal scaling (i.e. number of instances, ID of the subscribing instance, and dispatching method).

  • repeated Query query = 1

  • optional Sharding sharded_sub = 2

Search provides methods to query CloudVision using the Search service.

• rpc DeleteCustomSchema (CustomIndexSchemaDel, google.protobuf.Empty)

  router.proto:658

  for custom schema deletion This is alpha version of this api and doesn't synchronize across apiserver instances. apiserver restart is needed to get updated schema information from hbase.

• rpc Search (SearchRequest, stream NotificationBatch)

  router.proto:643

  you know, for search...

• rpc SearchSubscribe (SearchRequest, stream NotificationBatch)

  router.proto:646

  SearchSubscribe allows the client to request a live stream of updates based on client search request

• rpc SearchWithAggregation (SearchRequestWithAggr, AggrResponse)

  router.proto:648

  for search with aggregation

• rpc SearchWithAggregationStream (SearchRequestWithAggr, stream ByteStream)

  router.proto:652

  SearchWithAggregationStream sends the protobuf-serialized form of AggrResponse as in a stream of byteArrays. Receiver needs to append the "bytearrays", and protobuf-deserialize to obtain the result. Intended for messages exceeding the grpc size limit

• rpc SetCustomSchema (CustomIndexSchema, google.protobuf.Empty)

  router.proto:654

  for custom schema configuration

Used in: AggrResponse

• string key = 1

  key of the bucket

• uint64 count = 2

  count of matching document

• string aggr_name = 3

  aggr_name has name of the aggregation to which these buckets belong to this is useful when there are multiple top level aggregations Name is formed by prepending aggregation type to the field name for term aggreation with field ["a","b","c"] name would be termabc

• repeated AggrField fields = 11

  Numeric fields in this bucket

• repeated AggrBucket sub_buckets = 12

  Sub buckets within this bucket

Used in: AggrBucket, AggrResponse

• string name = 1

  Name of the field Name is formed by prepending aggregation type to the field name. SUM for field ["a","b","c"] would be SUMabc

• double value = 2

  Field value

Used as response type in: Alpha.SearchWithAggregation, Search.SearchWithAggregation

• repeated AggrBucket buckets = 1

  Buckets in aggregation response

• repeated AggrField fields = 2

  Numeric fields in aggregation response

Used in: SearchRequestWithAggr

• Aggregate.Type aggr_type = 1

  Type of aggregation being requested

• repeated string fields = 2

  Field on which aggregation operation needs to be applied Nested field like a.b.c needs to specified as ["a","b","c"]. It is not possible to combine fields only for aggregation operation. All the combinations need to be predecided at schema creation time only.

• SortType sort = 3

  For sorting on aggregated outcome. Applicable only for numeric aggregation.

• repeated Aggregate sub_aggrs = 4

  For specifying sub aggregation, like sum aggregation inside bucket aggregation.

• oneof Option

  Aggregation type specific additional options

  • term_options term = 21

    Option for term aggregation

  • hist_options hist = 26

    Option for histogram aggregation

Used in: Aggregate

• Invalid = 0

• TERM = 1

  Type of bucket aggregation where different buckets are created based on field values

• SUM = 2

  SUM aggregation where numeric summation is calculated on field values

• AVG = 3

  AVG aggregation where numeric average is calculated on field values

• MAX = 4

  MAX aggregation where numeric maximum is returned amongst the field values

• MIN = 5

  MIN aggregation where numeric minimum is returned amongst the field values

• HIST = 6

  histogram bucketing/aggregation on numeric field value

Used as response type in: Alpha.SearchWithAggregationStream, Search.SearchWithAggregationStream

• bytes message = 1

  Message is an array of bytes

Used as response type in: CertificateAuthority.Enroll, CertificateAuthority.Reenroll

• bytes crt = 1

Used as request type in: CertificateAuthority.Enroll, CertificateAuthority.Reenroll

• bytes csr = 1

Used in: CustomIndexSchema

• uint32 delete_after_days = 1

  delete_after_days is requested days after which data should get deleted

• uint32 inuse_delete_after_days = 2

  inuse_delete_after_days is what is actually in use. This is guaranteed to be atleast delete_after_days. delete_after_days may get modified based on index rotation policy by Cloudvision

Used as request type in: Alpha.SetCustomSchema, Search.SetCustomSchema

• optional Query query = 1

• repeated IndexField schema = 11

• optional CustomIndexOptions option = 21

Used as request type in: Alpha.DeleteCustomSchema, Search.DeleteCustomSchema

• optional Query query = 1

• bool delete_data = 11

  if delete_data is set, then all the data along with schema defintion is deleted. Data deletion is non recoverable operation !

Used in: CreateDatasetRequest, CreateSessionRequest, DatasetsRequest, DatasetsResponse, NotificationBatch, Query, SetPasswordRequest, SetPermissionRequest

• string type = 1

• string name = 2

• optional Dataset parent = 3

field name and filter value

Used in: SearchRequest

• string field = 1

• Filter.Operator op = 2

• optional Filter.Value value = 3

• repeated string nested_field = 11

  nested_field to support nested fields. a.b.c needs to specified as ["a","b","c"].

• repeated Filter sub_filters = 21

  sub_filters to specify complex predicate using logical operators. Complex predicates are specified as recursive definition of filters. Different levels are used to resemble abstract syntaxt tree like structure.

Used in: Value

• map<string, string> value = 1

Used in: Value

• repeated Value values = 1

Used in: Filter

• EQ = 0

  equal to

• NEQ = 1

  not equal to

• GT = 2

  greater than

• GE = 3

  greater than or equal to

• LT = 4

  less than

• LE = 5

  less than or equal to

• RE = 6

  regex

• NRE = 7

  not regex

• IN = 8

  in

• NIN = 9

  not in

• SUB = 10

  substring (prefix substring)

• WILDCARD = 11

  wildcard query (for custom index only)

• BITMASK = 16

  filter on bitmask bit(s) are set. Only applicable for integer fields

• LOGICALOR = 20

  Logical OR operator to be used between two filters

• LOGICALAND = 21

  Logical AND operator to be used between two filters

Value, normalized

Used in: Filter, MultiValue

• oneof kind

  • string str = 1

  • sint64 int = 2

  • uint64 uint = 3

  • double float = 4

  • bool b = 5

  • string ip = 6

  • string mac = 7

  • ComponentValue comp = 8

  • MultiValue multi = 9

Used as request type in: Auth.GetPermissionSet, RouterV1.Get

• repeated Query query = 1

  V1: NOTE: Query.Path.Type = REGEXP is not supported for GetRequest.

• uint64 start = 4

  (Optional) Lower bound (inclusive) of time range to retrieve, in nanoseconds since the Epoch. If `start` is set, the returned results will correspond to a time range query (from `start` to `end` or now, depending on whether `end` is set or not). If `start` is not set, this request will execute a "limit" query based on `versions`. The value 0 corresponds to an unset field.

• uint32 versions = 5

  (Optional) Maximum number of versions to retrieve. If `start` is set, this value is ignored. Otherwise, will return at most (`versions` + 1) past versions.

• uint64 end = 3

  (Optional) Upper bound (inclusive) of time range to retrieve, in nanoseconds since the Epoch. The value 0 corresponds to an unset field. Defaults to the largest valid timestamp.

• bool exact_range = 6

  (Optional) Whether to return the initial state of the requested paths at time `start`. The "initial state" is the set of notifications that fully specify the state of the path at that time. It includes all relevant notifications before (and up to) `start`. If `start` is not set, this is ignored.

• optional Sharding sharded_sub = 7

  (Optional) Number of client instances, id of the client making the request, and sharding method for the implementation of horizontal scaling. If not set, no filtering will happen.

• PathWildCardExpandType wildcard_expand_type = 8

  (Optional) Specifies wildcard expansion strategy when time range is specified. This option has no effect when time range (start/end) is not used.

Used in: IndexField

• INVALID = 0

• INTEGER = 1

• FLOAT = 2

• STRING = 3

• ARRAY = 11

• OBJECT = 12

Used in: CustomIndexSchema

• string name = 1

• IndexDataType type = 2

• repeated IndexField sub_field = 11

Used in: NotificationBatch

• optional google.protobuf.Timestamp timestamp = 1

  Timestamp in nanoseconds resolution.

• string path = 2

  Deprecated

• bool delete_all = 3

• repeated bytes deletes = 4

  encoded

• repeated Notification.Update updates = 5

• repeated bytes retracts = 6

  encoded

• repeated bytes path_elements = 7

  encoded

Used in: Notification, PublishRequest

• bytes key = 1

  encoded

• bytes value = 2

  encoded

Used as response type in: Alpha.Search, Alpha.SearchSubscribe, RouterV1.Get, RouterV1.GetAndSubscribe, RouterV1.Subscribe, Search.Search, Search.SearchSubscribe

Used as field type in: PublishRequest, SearchRequest

• string d = 1

  TODO: deprecate, for now just assume that type is "device"

• repeated Notification notifications = 2

• optional Dataset dataset = 3

• map<string, string> metadata = 4

• ReplicationState replicate = 5

Path defines a path matcher for a Query.

Used in: Query, SetPermissionRequest.PathPerm

• Path.Type type = 1

  V2: DEPRECATED

• string path = 2

  V2: DEPRECATED

• repeated bytes keys = 3

  Keys (V1: BSDN, V2: NEAT) can be optionally provided to narrow the set of notification keys included in the result.

• repeated bytes path_elements = 4

  PathElements (V2: NEAT) are a list of keys representing a path.

V2: DEPRECATED

Used in: Path

• EXACT = 0

  EXACT means that the "path" will be matched exactly

• REGEXP = 1

  REGEXP means that the "path" will be treated as regexp

Used in: GetAndSubscribeRequest, GetRequest

• WILDCARD_EXPAND_LEGACY = 0

  WILDCARD_EXPAND_LEGACY falls back to default behavior

• WILDCARD_EXPAND_LATEST = 1

  WILDCARD_EXPAND_LATEST means that only latest path pointers will be considered

• WILDCARD_EXPAND_EXACT_RANGE = 2

  WILDCARD_EXPAND_EXACT_RANGE means that path pointers satisfying given start/end will be considered

• WILDCARD_EXPAND_RELAXED_RANGE = 3

  WILDCARD_EXPAND_RELAXED_RANGE means that path pointers satisfying 0 to end will be considered

Used in: CustomIndexSchema, CustomIndexSchemaDel, GetAndSubscribeRequest, GetRequest, SearchRequest, SubscribeRequest

• optional Dataset dataset = 1

• repeated Path paths = 2

ReplicationState represent the level of replication we want to apply on a notification. By default there is no replication. If CACHE_REPLICATE is used then that notification will be replicated to all the other clusters connected to the one where that notification is currently published, and the caches of the components running in these clusters will be synced together. REPLICATE is like CACHE_REPLICATE but without the cache synchronization.

Used in: NotificationBatch

• NO_REPLICATION = 0

• REPLICATE = 1

• CACHE_REPLICATE = 2

Metadata, usually returned with the first response. But it could return some metadata on any subsequent response, including the last one. The message can contain partial data (for instance, columns will be only in the first answer)

Used in: SQLResponse

• repeated string columns = 1

  Name of the columns for the query result returned Returned only in the first response message.

Used in: SQLResponse

• repeated bytes values = 1

  NEAT encoded values for one row

Used as request type in: Alpha.Search, Alpha.SearchSubscribe, Search.Search, Search.SearchSubscribe

Used as field type in: SearchRequestWithAggr

• string search = 1

  A search string

• SearchRequest.Type search_type = 2

  The type of value that's being searched.

• repeated Query query = 3

  Filter on specific datasets and paths.

• uint64 start = 4

  Start and end time of the notifications returned, inclusive.

• uint64 end = 5

• bool count_only = 6

  Return only the number of results

• repeated Filter key_filters = 7

  Complex fields search These are applicable for CUSTOM type as well

• repeated Filter value_filters = 8

• repeated Sort sort = 9

  Sort criteria

• bool exact_range = 10

  (Optional) Whether to return the initial state of the requested paths at time `start`. The "initial state" is the set of notifications that fully specify the state of the path at that time. It includes all relevant notifications before (and up to) `start`. If `start` is not set, this is ignored.

• optional Sharding sharded_sub = 11

  (Optional) Set NumParentShards field >= 1 to enable multi-tenant search

• bool exact_term = 12

  (Optional) whether search exact term or not, useful for string search

• uint32 result_size = 21

  Total number of notification to get back in the response

• uint32 offset = 22

  Numeric offset as starting point for next page of result

• optional NotificationBatch meta = 23

  cursor based pagination

Used in: SearchRequest

• STRING = 0

  STRING means that Key/Value string fields will be searched

• MAC = 1

  MAC means that Key/Value mac fields will be searched

• IP = 2

  IP means that Key/Value ip fields will be searched

• COMPLEX = 3

  COMPLEX means that a complex key or value will be searched

• CUSTOM = 4

  CUSTOM means that custom schema is used. Path prefix is mandatory in query.

Used as request type in: Alpha.SearchWithAggregation, Alpha.SearchWithAggregationStream, Search.SearchWithAggregation, Search.SearchWithAggregationStream

• optional SearchRequest search = 1

  Search criteria within aggregation request

• repeated Aggregate aggrs = 2

  Aggregation clauses

SetPermissionRequest encompasses requests to create/update/delete permissions of each type. A PATH_PERMISSION must contain a set of path<->permission tuples. The INHERIT_PERMISSION only requires the @dataset and @other args. ADMIN_PERMISSION reuses @path but only looks at the @perm field of the first element. Updates require the the client to pass `oldPerm` to avoid race conditions.

Used as request type in: Auth.SetPermission

Used as field type in: PermissionSet

• SetPermissionRequest.Type type = 1

• optional Dataset dataset = 2

• optional Dataset other = 3

• repeated SetPermissionRequest.PathPerm pathPerms = 6

Used in: SetPermissionRequest

• optional Path path = 1

• PathPerm.Perm newPerm = 2

• PathPerm.Perm currentPerm = 3

• bool exactMatch = 4

Used in: PathPerm

• NULL_PERM_VALUE = 0

• READ_PERM = 1

• WRITE_PERM = 2

• READ_WRITE_PERM = 3

Used in: SetPermissionRequest

• PATH_PERMISSION = 0

• INHERIT_PERMISSION = 1

• UNINHERIT_PERMISSION = 2

• ADMIN_PERMISSION = 3

• SET_ROLE_PERMISSION = 4

• REMOVE_ROLE_PERMISSION = 5

Sharding contains the information for horizontal scaling of subscriptions and get requests. Multitenancy, i.e., matching notifications across multiple parent datasets, is enabled by setting numParentShards >= 1. Dataset/Path sharding must be configured orthogonally to Parent sharding. This makes it possible to map an entire org to a single client instance. Examples: {numShards = 1, numParentShards = M} => distributes notifs among M shards preserving org grouping {numShards = N, numParentShards = 1} => distributes among N shards without preserving org grouping {numShards = N, numParentShards = M} => means that there will be N*M shards in total where the notification from a single org will be sharded into N bins based on the ShardingFunction.

Used in: DatasetsRequest, GetAndSubscribeRequest, GetRequest, SearchRequest, SubscribeRequest

• uint32 shard = 1

  shard is the unique ID of the client instance for horizontal scaling

• uint32 numShards = 2

  numShards is the number of instances of the client for horizontal scaling

• ShardingFunction shardingFunc = 3

  shardingFunc defines which sharding function to use for horizontal scaling

• uint32 parentShard = 4

  parentShard is the unique ID of the client instance if multitenant is enabled

• uint32 numParentShards = 5

  numParentShards is the number of client instances for parent dataset sharding

Used in: Sharding

• DATASETANDPATH = 0

  DATASETANDPATH shards by the association of dataset and path

• DATASET = 1

  DATASET shards by dataset

• PATH = 2

  PATH shards by path

Used in: SearchRequest

• repeated string fields = 1

  fields to support nesting. a.b.c needs to specified as ["a","b","c"].

• SortType type = 2

  type of sort

Used in: Aggregate, Sort

• SORT_INVALID = 0

• DESC = 1

  Descending order sort

• ASC = 2

  Ascending order sort

Used in: Aggregate

• uint64 interval = 1

  mandatory

• bool allow_empty = 2

  optional

• optional hist_options.bounds extended_bounds = 3

  specify bounding box beyond min/max returned from query, together with allow_empty

Used in: hist_options

• int64 min = 1

• int64 max = 2

Used in: Aggregate

• uint32 num_of_buckets = 1

  For controlling maximum number of buckets

• term_options.Type field_type = 2

For providing type hint

Used in: term_options

• UNSPECIFIED = 0

• BOOL = 1

• DOUBLE = 2

• LONG = 3

• STRING = 4

• COMPONENT = 5

• IP = 6

• MAC = 7