Built-in Endpoints

This page describes the REST API endpoints accessible on a TigerGraph server.

Note for users of TigerGraph Cloud

For all TigerGraph Cloud solutions created on or after June 20, 2022, ports 14240 and 9000 are no longer exposed. This also applies to previously created cloud solutions that are resized on or after June 20, 2022. Use port 443 instead for API endpoint connectivity on cloud solutions:

  • Use API endpoint 443/restpp/<endpoint> for RESTPP endpoints that were on port 9000

  • Use API endpoint 443/<endpoint> for endpoints that were on port 14240

For all TigerGraph Server solutions not on TigerGraph Cloud, or for TigerGraph Cloud solutions created before June 20, 2022, ports 9000 and 14240 remain correct.

System Utilities

Echo

GET /echo and POST /echo

These endpoints are simple diagnostic utilities, which respond with the following message if the RESTPP server is up and running.

Sample request:

GET echo/ Request and Response
curl -X GET "http://localhost:9000/echo"
{
    "error": false,
    "message": "Hello GSQL"
}

POST /echo has the same response as GET /echo.

Parameters

Name Required Description

sleep

No

Integer that indicates the number of seconds for which the response will be delayed.

Health check (public)

This endpoint performs a simple server health check. It listens on port 14240 and does not require authentication. If you ping it and the server is running, it will respond with the message "pong".

Endpoint:

GET /api/ping

Sample request:

curl 'http://localhost:14240/api/ping'
{
  "error": "false",
  "message": "pong",
  "results": {}
}

Parameters:

No parameters.

List all endpoints

GET /endpoints/{graph_name}

This endpoint returns a list of the installed endpoints and their parameters. There are three types of endpoints:

  • Built-in endpoints which are preinstalled in the TigerGraph system

  • Dynamic endpoints which are generated when compiling GSQL queries

  • Static endpoints which are user-installed endpoints

To include one or more of the endpoint types in the output, include the endpoint type in the parameter query string and set its value to true. If no type parameters are provided, all endpoints are returned.

Example: Report on all built-in endpoints
curl -X GET "http://localhost:9000/endpoints?builtin=true" | jq .

Response

There are over a dozen built-in endpoints, and some have several parameters, so the formatted JSON output of all built-in endpoints is over 300 lines long. It is listed in full in Appendix A. Below is a small excerpt of the output:

Subset of GET /endpoints output
    "GET /endpoints": {
        "parameters": {
            "builtin": {
                "default": "false",
                "max_count": 1,
                "min_count": 0,
                "type": "BOOL"
            },
            "dynamic": {
                "default": "false",
                "max_count": 1,
                "min_count": 0,
                "type": "BOOL"
            },
            "static": {
                "default": "false",
                "max_count": 1,
                "min_count": 0,
                "type": "BOOL"
            }
        }
    }

Parameters

Name Required Description

builtin

No

Takes a boolean value. Returns built-in endpoints if true.

dynamic

No

Takes a boolean value. Returns dynamic endpoints if true.

static

No

Takes a boolean value. Returns user-installed endpoints if true.

Show component versions

GET /version

This endpoint returns the Git versions of all components of the system.

  • Request

  • Response

curl -X GET "http://localhost:9000/version"
{"error":"false", "message":"TigerGraph RESTPP:
 --- Version ---
product              release_2.6.0_05-09-2020 ab1e3d0da6237c27468d6cabb90900119d63759d  2020-04-15 15:46:29 -0700
olgp                 release_2.6.0_05-09-2020 046c745088106b69920b9bdb3bd15969de409e92  2020-05-01 19:10:27 -0700
topology             release_2.6.0_05-09-2020 c028af100117f2051b619436c3aa4febc810bf36  2020-04-22 08:44:07 -0700
gpe                  release_2.6.0_05-09-2020 34b9e86ef7b5fdaa106637e7db1d8a9e080a0aa2  2020-04-19 09:42:59 -0700
gse                  release_2.6.0_05-09-2020 ed2c2351357aa9077fa4dee7ea7a01f8ad2f7585  2020-05-11 01:18:54 -0700
third_party          release_2.6.0_05-09-2020 4bce6990bae5be2b91e9201693ceb66341d3f204  2020-04-19 09:42:56 -0700
utility              release_2.6.0_05-09-2020 2ce197d3edb3557bdd66ed1a4194309908d6197e  2020-04-20 21:19:34 -0700
realtime             release_2.6.0_05-09-2020 52a82b454437c73b47d846acd5803ab0d9f54a45  2020-04-22 08:44:11 -0700
er                   release_2.6.0_05-09-2020 a3e6cb7606fb74984c75cae9bbd4d2112fdbf73a  2020-05-01 19:10:33 -0700
gle                  release_2.6.0_05-09-2020 d8bdbd1cf346e181aa9a317c704dd7b3b11b4658  2020-05-06 00:51:04 -0700
bigtest              release_2.6.0_05-09-2020 2f64c47b7a5ac1834ead9a22eef8d42241117853  2019-12-12 01:31:35 -0800
document             release_2.6.0_05-09-2020 6327094bd76b2dbc8f4625108d547827344b5091  2019-12-13 16:30:13 -0800
glive                release_2.6.0_05-09-2020 93f61ea06fe42759c808fc58ff6245c9954d5447  2020-02-05 22:40:24 -0800
gap                  release_2.6.0_05-09-2020 e798efb595545bf91c449034566857c41f52449a  2020-04-29 22:47:26 -0700
gst                  release_2.6.0_05-09-2020 1b695c02f277efad0ddfb2deab710ae0158409da  2020-04-29 22:47:32 -0700
gus                  release_2.6.0_05-09-2020 eee784502b5387844e462305bae419954784da6f  2020-04-29 22:47:20 -0700
blue_features        release_2.6.0_05-09-2020 5d7a4e8d806519f529274b331496d3bc78f01990  2020-04-15 15:46:38 -0700
blue_commons         release_2.6.0_05-09-2020 432763afc49bf986aed4731e50254243d3665bc3  2019-07-30 03:34:46 -0700
"}

Parameters

This endpoint does not take any parameters.

Monitor system metrics

GET /ts3/api/datapoints

TigerGraph System State Service (TS3) is TigerGraph’s managed monitoring service that collects system metrics and events. Many TigerGraph components report metrics such as CPU usage, memory usage, disk usage, and network traffic to TS3 at regular intervals. This endpoint listens on port 14240 instead of port 9000.

You can use this endpoint to read from TS3. You can filter for the data points you need as follows:

  • time: when, from, and to

  • component: who

  • metric: what

  • location: where

Visualization of such metrics are available in Admin Portal - Dashboard - Cluster Monitoring.

On a TigerGraph cluster, this endpoint is only present on the m1 node.

Parameters

Name Required Description

from

No

Epoch timestamp that indicates the start of the time filter. Only data points reported after the timestamp will be included in the return results.

to

No

Epoch timestamp that indicates the end of the time filter. Only data points reported before the timestamp will be included in the return results.

latest

No

Number of latest data points to return. If provided, the endpoint will return the latest data points that satisfy the what, who and where filters and ignore other time-related filters.

what

No

Name of the metric to filter for. Possible values are:

  • cpu: Percentage of CPU usage by component

  • mem: Memory usage in megabytes by component

  • diskspace: Disk usage in megabytes by directory

  • network: Network traffic in bytes since the service started

  • qps: Number of requests per second by endpoint

  • servicestate: The state of the service, either online 1 or offline 0

  • connection: Number of open TCP connections

who

No

Name of the component that reported the datapoint

where

No

Name of the node that the datapoint is reported for

Sample requests

The sample request below uses three filters: from, to and what.

from and to specify a timeframe with start and end times of 1618957536 and 1619023346.

The what filter specifies that the response should only include CPU information:

  • Request

  • Response

curl -X GET
"https://crunch.i.tgcloud.io:14240/ts3/api/datapoints?from=1618957536&to=1619023346&what=cpu"
# Three data points returned
[
  {
    "detail": 0,  # GPE is using 0 percent CPU
    "when": 1619023346,
    "where": "m1",
    "who": "GPE_1#1",
    "what": "cpu"
  },
  {
    "detail": 0,
    "when": 1619023346,
    "where": "m1",
    "who": "GSE_1#1",
    "what": "cpu"
  },
  {
    "detail": 0,
    "when": 1619023346,
    "where": "m1",
    "who": "RESTPP#1",
    "what": "cpu"
  }
]

In the below example, what and latest are used. The request asks for the 10 latest data points regarding memory usage:

  • Request

  • Response

curl -X GET "https://crunch.i.tgcloud.io:14240/ts3/api/datapoints?what=mem&latest=10"
[
  {
    "detail": 159,
    "when": 1620076473,
    "where": "m1",
    "who": "RESTPP#1",
    "what": "mem"
  },
  {
    "detail": 211,
    "when": 1620076533,
    "where": "m1",
    "who": "GPE_1#1",
    "what": "mem"
  },
  {
    "detail": 436,
    "when": 1620076533,
    "where": "m1",
    "who": "GSE_1#1",
    "what": "mem"
  },
  {
    "detail": 159,
    "when": 1620076533,
    "where": "m1",
    "who": "RESTPP#1",
    "what": "mem"
  },
  {
    "detail": 211,
    "when": 1620076593,
    "where": "m1",
    "who": "GPE_1#1",
    "what": "mem"
  },
  {
    "detail": 436,
    "when": 1620076593,
    "where": "m1",
    "who": "GSE_1#1",
    "what": "mem"
  },
  {
    "detail": 159,
    "when": 1620076593,
    "where": "m1",
    "who": "RESTPP#1",
    "what": "mem"
  },
  {
    "detail": 210,
    "when": 1620076653,
    "where": "m1",
    "who": "GPE_1#1",
    "what": "mem"
  },
  {
    "detail": 436,
    "when": 1620076653,
    "where": "m1",
    "who": "GSE_1#1",
    "what": "mem"
  },
  {
    "detail": 159,
    "when": 1620076653,
    "where": "m1",
    "who": "RESTPP#1",
    "what": "mem"
  }
]

Monitor system metrics (OpenMetrics)

GET :9167/metrics

You can monitor system metrics regarding CPU usage, memory usage, network usage, diskspace, service status, or engine QPS data in OpenMetrics format, allowing you to integrate this endpoint with observability platforms such as Datadog or Prometheus. This endpoint exists on port 9167 by default.

This endpoint listens on the same port as the informant service in TigerGraph. If you change the informant port to a different one, this endpoint listens on that port instead.

Every call to this endpoint retrieves all metrics. Metrics are updated every 60 seconds by default. This interval can be adjusted with the System.Metrics parameters.

This endpoint does not require any privilege.

Parameters

None.

Sample request

  • Request

  • Response

$ curl http://localhost:9167/metrics
# HELP tigergraph_cpu_usage Percentage of CPU usage by component. Service name empty means usage of the whole node
# TYPE tigergraph_cpu_usage gauge
tigergraph_cpu_usage{host_id="m1",service_name=""} 62.04006576538086
tigergraph_cpu_usage{host_id="m1",service_name="ADMIN"} 0.4452745020389557
tigergraph_cpu_usage{host_id="m1",service_name="CTRL"} 0.3366197347640991
tigergraph_cpu_usage{host_id="m1",service_name="DICT"} 0.13882260024547577
tigergraph_cpu_usage{host_id="m1",service_name="ETCD"} 1.021435022354126
tigergraph_cpu_usage{host_id="m1",service_name="GPE"} 7.542555809020996
tigergraph_cpu_usage{host_id="m1",service_name="GSE"} 45.86964416503906
tigergraph_cpu_usage{host_id="m1",service_name="GSQL"} 1.0639517307281494
tigergraph_cpu_usage{host_id="m1",service_name="GUI"} 0.010601896792650223
tigergraph_cpu_usage{host_id="m1",service_name="IFM"} 1.0899430513381958
tigergraph_cpu_usage{host_id="m1",service_name="KAFKA"} 56.06120681762695
tigergraph_cpu_usage{host_id="m1",service_name="KAFKACONN"} 1.8662768602371216
tigergraph_cpu_usage{host_id="m1",service_name="KAFKASTRM-LL"} 0.9830034971237183
tigergraph_cpu_usage{host_id="m1",service_name="NGINX"} 0
tigergraph_cpu_usage{host_id="m1",service_name="RESTPP"} 13.063582420349121
tigergraph_cpu_usage{host_id="m1",service_name="TS3"} 0.1956222653388977
tigergraph_cpu_usage{host_id="m1",service_name="TS3SERV"} 0.1768123060464859
tigergraph_cpu_usage{host_id="m1",service_name="ZK"} 0.25204044580459595
# HELP tigergraph_diskspace_free Free Disk space in megabytes by directory.
# TYPE tigergraph_diskspace_free gauge
tigergraph_diskspace_free{host_id="m1",mount_point="/",path="/home/tigergraph",path_name="Home"} 188097.546875
tigergraph_diskspace_free{host_id="m1",mount_point="/",path="/home/tigergraph/tigergraph/data/gstore",path_name="Gstore"} 188097.546875
tigergraph_diskspace_free{host_id="m1",mount_point="/",path="/home/tigergraph/tigergraph/data/kafka",path_name="Kafka"} 188097.546875
tigergraph_diskspace_free{host_id="m1",mount_point="/",path="/home/tigergraph/tigergraph/log",path_name="Log"} 188097.546875
# HELP tigergraph_diskspace_usage Disk usage in megabytes by directory.
# TYPE tigergraph_diskspace_usage gauge
tigergraph_diskspace_usage{host_id="m1",mount_point="/",path="/home/tigergraph",path_name="Home"} 7107.47509765625
tigergraph_diskspace_usage{host_id="m1",mount_point="/",path="/home/tigergraph/tigergraph/data/gstore",path_name="Gstore"} 579.0476684570312
tigergraph_diskspace_usage{host_id="m1",mount_point="/",path="/home/tigergraph/tigergraph/data/kafka",path_name="Kafka"} 1164.056640625
tigergraph_diskspace_usage{host_id="m1",mount_point="/",path="/home/tigergraph/tigergraph/log",path_name="Log"} 255.07992553710938
# HELP tigergraph_memory_usage Memory usage in megabytes by component. Service name empty means usage of the whole node
# TYPE tigergraph_memory_usage gauge
tigergraph_memory_usage{host_id="m1",service_name=""} 6452
tigergraph_memory_usage{host_id="m1",service_name="ADMIN"} 44
tigergraph_memory_usage{host_id="m1",service_name="CTRL"} 33
tigergraph_memory_usage{host_id="m1",service_name="DICT"} 35
tigergraph_memory_usage{host_id="m1",service_name="ETCD"} 22
tigergraph_memory_usage{host_id="m1",service_name="GPE"} 682
tigergraph_memory_usage{host_id="m1",service_name="GSE"} 1129
tigergraph_memory_usage{host_id="m1",service_name="GSQL"} 287
tigergraph_memory_usage{host_id="m1",service_name="GUI"} 28
tigergraph_memory_usage{host_id="m1",service_name="IFM"} 33
tigergraph_memory_usage{host_id="m1",service_name="KAFKA"} 646
tigergraph_memory_usage{host_id="m1",service_name="KAFKACONN"} 2182
tigergraph_memory_usage{host_id="m1",service_name="KAFKASTRM-LL"} 401
tigergraph_memory_usage{host_id="m1",service_name="NGINX"} 2
tigergraph_memory_usage{host_id="m1",service_name="RESTPP"} 295
tigergraph_memory_usage{host_id="m1",service_name="TS3"} 20
tigergraph_memory_usage{host_id="m1",service_name="TS3SERV"} 21
tigergraph_memory_usage{host_id="m1",service_name="ZK"} 116
# HELP tigergraph_network_connections Number of open TCP connections.
# TYPE tigergraph_network_connections gauge
tigergraph_network_connections{host_id="m1",ip="10.128.0.203"} 477
# HELP tigergraph_network_traffic Network traffic in bytes since the service started.
# TYPE tigergraph_network_traffic gauge
tigergraph_network_traffic{direction="incoming",host_id="m1",ip="10.128.0.203"} 4.579066608e+09
tigergraph_network_traffic{direction="outgoing",host_id="m1",ip="10.128.0.203"} 4.579066608e+09
# HELP tigergraph_service_status Tigergraph service status. Partiton or replica empty means no partition or replica for that service.
# TYPE tigergraph_service_status gauge
tigergraph_service_status{partition="",replica="1",service_name="ADMIN"} 6
tigergraph_service_status{partition="",replica="1",service_name="CTRL"} 6
tigergraph_service_status{partition="",replica="1",service_name="DICT"} 6
tigergraph_service_status{partition="",replica="1",service_name="ETCD"} 6
tigergraph_service_status{partition="",replica="1",service_name="GSQL"} 6
tigergraph_service_status{partition="",replica="1",service_name="GUI"} 6
tigergraph_service_status{partition="",replica="1",service_name="IFM"} 6
tigergraph_service_status{partition="",replica="1",service_name="KAFKA"} 6
tigergraph_service_status{partition="",replica="1",service_name="KAFKACONN"} 6
tigergraph_service_status{partition="",replica="1",service_name="NGINX"} 6
tigergraph_service_status{partition="",replica="1",service_name="RESTPP"} 6
tigergraph_service_status{partition="",replica="1",service_name="TS3SERV"} 6
tigergraph_service_status{partition="",replica="1",service_name="ZK"} 6
tigergraph_service_status{partition="1",replica="",service_name="EXE"} 6
tigergraph_service_status{partition="1",replica="",service_name="KAFKASTRM-LL"} 6
tigergraph_service_status{partition="1",replica="",service_name="TS3"} 6
tigergraph_service_status{partition="1",replica="1",service_name="GPE"} 6
tigergraph_service_status{partition="1",replica="1",service_name="GSE"} 6

Show query performance

GET /statistics/{graph_name}

This endpoint returns real-time query performance statistics over the given time period, as specified by the seconds parameter. Seconds are measured up to 60, so the seconds parameter must be a positive integer less than or equal to 60.

Sample request:

The return object is a hash of the endpoints and their performance data.

This example shows two endpoints (/graph/vertex and /statistics) called during the past 60 seconds.

  • Request

  • Response

curl -X GET "http://localhost:9000/statistics/poc_graph?seconds=60" | jq '.'
{
  "GET /graph/vertices/{vertex_type}/{vertex_id}": {
    "CompletedRequests": 8,
    "QPS": 0.08,
    "TimeoutRequests": 0,
    "AverageLatency": 130,
    "MaxLatency": 133,
    "MinLatency": 128,
    "LatencyPercentile": [
      200,
      200,
      200,
      200,
      200,
      200,
      200,
      200,
      200,
      200
    ]
  },
  "GET /statistics": {
    "CompletedRequests": 4226,
    "QPS": 42.26,
    "TimeoutRequests": 0,
    "AverageLatency": 2,
    "MaxLatency": 125,
    "MinLatency": 0,
    "LatencyPercentile": [
      10,
      10,
      10,
      10,
      10,
      10,
      10,
      10,
      10,
      200
    ]
  }
}

Each endpoint has the following attributes:

  • CompletedRequests - the number of completed requests.

  • QPS - query per second.

  • TimeoutRequests - the number of requests not returning before the system-configured timeout limit. Timeout requests are not included in the calculation of QPS.

  • AverageLatency - the average latency of completed requests.

  • MaxLatency - the maximum latency of completed requests.

  • MinLatency - the minimum latency of completed requests.

  • LatencyPercentile - The latency distribution. The number of elements in this array depends on the segments parameter of this endpoint whose default value is 10, meaning the percentile range 0-100% will be divided into ten equal segments: 0%-10%, 11%-20%, etc. Segments must be between 1 and 100.

If there is no query sent in the past given seconds, an empty json will be returned.

Parameters

Name Required Description

seconds

Yes

Positive integer less than 60 that indicates how many seconds back from the current time the statistics report will cover.

segments

No

Integer that indicates the number of segments that the LatencyPercentile array in the response will be split into. The value for this endpoint must be between 1 and 100. The default value is 10.

Show service status

POST :14240/informant/current-service-status

This endpoint returns the status of the TigerGraph services specified in the request.

Parameters

No URL parameters.

Request body

The endpoint requires a request body in the following format:

{
    "ServiceDescriptors": [ (1)
        {
            "ServiceName": <service_name>, (2)
            "Partition": <partition_number>, (3)
            "Replica": <replica_number> (4)
        }
        ... (5)
    ]
}
1 ServiceDescriptors is a required field. It is a list of objects with keys ServiceName, and optionally Partition and Replica.
2 Required field. The name of the service to return status on. Below are the accepted values for the field and their corresponding services:
  • "GPE": GPE

  • "GSE": GSE

  • "RESTPP": RESTPP

  • "GSQL": GSQL

  • "IFM": INFORMANT

  • "GUI": GUI

  • "CTRL": CONTROLLER

  • "KAFKA": KAFKA

  • "ETCD": ETCD

  • "ZK": ZOOKEEPER

  • "NGINX": NGINX

  • "TS3": TS3

  • "TS3SERV": TS3SERV

  • "DICT": DICT

  • "ADMIN": ADMIN

3 Optional. Number of the partition to request service status on. If not provided, the response will contain service status on all partitions.
4 Optional. Number of the replica to request service status on. If not provided, the response will contain service status on all replicas.
5 You can supply more than one service descriptor. The response from the endpoint will contain all service status requested.

Return value

The return value contains the status for every service descriptor in the request. The most important information is in the ServiceStatus and ProcessState fields:

{
  "ServiceStatusEvents": [
    {
      "EventMeta": {
        "Targets": [
          {
            "ServiceName": "IFM"
          }
        ],
        "EventId": "154e8f53716b403eb02af19d863745c6",
        "SpanId": "ServiceStatusSelfReport",
        "TimestampNS": "1635841759229416893",
        "Source": {
          "ServiceName": "GPE",
          "Replica": 1,
          "Partition": 2
        }
      },
      "ServiceDescriptor": { (1)
        "ServiceName": "GPE",
        "Replica": 1,
        "Partition": 2
      },
      "ServiceStatus": "Online", (2)
      "ProcessState": "Running" (3)
    }
  ]
}
1 The service descriptor for the status being returned.
2 The last recorded status of the service.
3 The last recorded state of the Linux process behind the service. It’s possible for a process to be running without being served or able to respond to requests. For example, when TigerGraph is starting up, GPE is in the "Warmup" state and cannot respond to requests, but the process is still running.

Example

  • Request

  • Response

curl -X POST http://localhost:14240/informant/current-service-status -d '{ "ServiceDescriptors":  [{ "ServiceName": "gpe","Partition": 2, "Replica": 1}]}' | jq
{
  "ServiceStatusEvents": [
    {
      "EventMeta": {
        "Targets": [
          {
            "ServiceName": "IFM"
          }
        ],
        "EventId": "154e8f53716b403eb02af19d863745c6",
        "SpanId": "ServiceStatusSelfReport",
        "TimestampNS": "1635841759229416893",
        "Source": {
          "ServiceName": "GPE",
          "Replica": 1,
          "Partition": 2
        }
      },
      "ServiceDescriptor": {
        "ServiceName": "GPE",
        "Replica": 1,
        "Partition": 2
      },
      "ServiceStatus": "Online",
      "ProcessState": "Running"
    }
  ]
}

Rebuild graph engine

GET /rebuildnow/{graph_name} or

POST /rebuildnow/{graph_name}

Required privilege: Global-level UPDATE_DATA

When new data is being loaded into the graph (such as new vertices or edges), data is first stored in memory before it is saved to disk permanently. TigerGraph runs a rebuild of the Graph Processing Engine (GPE) to commit the data in memory to disk every 30 seconds, but you can also call this endpoint to trigger a rebuild immediately.

Parameters

Name Required Description

threadnum

No

Number of threads used to execute the rebuild. If not specified, the number specified in the .tg.cfg file ("RebuildThreadNumber") in the home directory of the server on which TigerGraph is running will be used; it is set to 3 by default. The maximum value for this parameter is the number of vCPUs per node in your distributed system. If you are running a single-node server, the maximum is the number of vCPUs on that node. You can run lscpu in your Linux server command line and look in the CPU(s) column to view the number of vCPUs.

vertextype

No

Vertex type to perform the rebuild for. If not provided, the rebuild will be run for all the vertex types.

segid

No

Segment ID of the segments to rebuild. If not provided, all segments will be rebuilt. In general, it is recommended not to provide this parameter and rebuild all segments.

path

No

Path to save the summary of the rebuild to. If not provided, the default path is /tmp/rebuildnow

force

No

Boolean value that indicates whether to perform rebuilds for segments for which there are no records of new data. Normally, a rebuild would skip such segments, but if force is set true, the segments will not be skipped.

Example

curl -X GET 'http://localhost:9000/rebuildnow/social'

# JSON response
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "RebuildNow finished, please check details in the folder: /tmp/rebuildnow",
  "results": [],
  "code": "REST-0000"
}

# Example summary file
cat finished.summary.txt

[SELECTED]	Segment id: 106, vertextype: 0, vertexsubtypeid: 0, vertexcount: 187732, edgecount: 563196, deletevertexcount: 0, postqueue_pos: 16344, transaction id: 16344, rebuild ts: 1573106412990
[SKIPPED]	Segment id: 6, vertextype: 0, vertexsubtypeid: 0, vertexcount: 85732, edgecount: 3106, deletevertexcount: 0, postqueue_pos: 16344, transaction id: 16344, rebuild ts: 1573106412900

Check deleted vertices

GET /deleted_vertex_check

Required privilege: Global-level DELETE_DATA

In certain rare cases, TigerGraph’s Graph Processing Engine (GPE) and Graph Storage Engine (GSE) might be out of sync on vertex deletion information. When this happens, some vertices might exist on one of the components, but not the other. Even though these errors are exceedingly rare, TigerGraph provides an endpoint that allows you to check the deleted vertices on GSE and GPE to see if they are out of sync.

The check passes if there are no discrepancies between the GSE and GPE in terms of deleted vertices. If there is a discrepancy, the check fails and the return result will contain the IDs of the deleted vertices that are not synced properly. If you are running TigerGraph on a distributed cluster, the check will be performed on each node of the cluster, and the endpoint will return a list containing the results of the check for every node.

Parameters

Name Required Description

threadnum

No

Integer that indicates the number of threads used to execute the deleted vertex check jobs. This parameter is optional and the default value is 6 if none is provided

segid

No

IDs of segments to perform the deleted vertex check for. If none is provided, the check will be performed on all segments.

vertextype

No

Vertex types to perform the deleted vertex check for. If none is provided, the check will be performed on all vertex types.

verbose

No

Integer that indicates the level of detail in the return results. Here is a list of accepted values and their corresponding level of detail:

  • 0 (default) : Only return whether the check passed and the list of unsynced vertex IDs

  • 1: In addition to the previous level, also return vertex count information

  • 2: In addition to the previous level, return vertex count information for every segment

  • 4: In addition to the previous level, also return the IDs of deleted vertices for every segment

log

No

Integer that indicates the log level of the deleted vertex check. This log is not returned in the endpoint’s HTTP response, but is printed to the logs of the GPE component at /tigergraph/log/gpe/log.INFO

  • 0 (default): Report brief log for the check as a whole

  • 1: Report logs for each segment

  • 2: Report additional logs on the obtained deleted ID list

Example:

# Passing check performed on a single-node database
curl -X GET "http://localhost:9000/deleted_vertex_check?threadnum=10&verbose=0" |jq .

{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "check passed",
  "results": [
    {
      "GPE": "GPE_1_1",
      "PassCheck": true,
      "UnSyncList": []
    }
  ],
  "code": "REST-0000"
}

# Failed check performed on a distributed cluster

curl -X GET 'http://localhost:9000/deleted_vertex_check?threadnum=10&verbose=0&vertextype=region' |jq .
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "check failed",
  "results": [
    {
      "GPE": "GPE_2_1",
      "PassCheck": false,
      "UnSyncList": [
        {
          "Segid": 193,
          "IsRemote": false,
          "VertexType": "region",
          "GPEDelHash": 7013042118817697000,
          "IDSDelHash": 202375168
        }
      ]
    },
    {
      "GPE": "GPE_3_1",
      "PassCheck": false,
      "UnSyncList": [
        {
          "Segid": 193,
          "IsRemote": true,
          "VertexType": "region",
          "GPEDelHash": 7013042118817697000,
          "IDSDelHash": 202375168
        }
      ]
    },
    {
      "GPE": "GPE_1_1",
      "PassCheck": false,
      "UnSyncList": [
        {
          "Segid": 193,
          "IsRemote": true,
          "VertexType": "region",
          "GPEDelHash": 7013042118817697000,
          "IDSDelHash": 202375168
        }
      ]
    }
  ],
  "code": "REST-0000"
}

Authentication

The endpoints in this subsection allow users to create, refresh and delete authentication tokens for requests made to the REST++ server. These endpoints only exist when user authentication is enabled on RESTPP endpoints.

Request a token

POST /requesttoken

This endpoint returns a token used to authenticate RESTPP requests.

If authentication is enabled on RESTPP endpoints, a token needs to be included in the request header for all requests sent to the RESTPP server. A user can generate a token using either

  • A secret, which is a random string generated in GSQL (see User Credentials)

  • Their username and password.

    • The request also needs to specify the graph for which the token is valid if it uses username and password to generate the token.

Request body

The endpoint expects a JSON request body in the following format:

{
    "secret": <secret_value>, (1)
    "graph": <graph_name>, (2)
    "lifetime": <token_lifetime> (3)
}
1 User’s secret to generate the token. Required if the request body does not supply graph.
2 Name of the graph that the token is valid for. Required if the request body does not supply secret.
3 Period of time for which the token is valid measured in seconds. The default value is about 2.6 million (about a month).

Sample request

The responses are slightly different between requests made with secrets and username-password pair.

  • With secret

  • Response

curl -X POST http://localhost:9000/requesttoken \
     -d '{"secret":"jiokmfqqfu2f95qs6ug85o89rpkneib3", "graph":"MyGraph", "lifetime":"100000"}'
{
  "code": "REST-0000",
  "expiration": 1616042814,
  "error": false,
  "message": "Generate new token successfully.\nWarning: Tigergraph Support cannot restore access to secrets/tokens for security reasons. Please save your secret/token and keep it safe and accessible.",
  "token": "tohvf6khjqju8jf0r0l1cohhlm8gi5fq"
}
  • With username-password pair

  • Response

curl --user example_username:example_password \
     -X POST http://localhost:9000/requesttoken \
     -d '{"graph": "example_graph"}'
{
  "code": "REST-0000",
  "expiration": 1648410519,
  "error": false,
  "message": "Generate new token successfully.\nWarning: Tigergraph Support cannot restore access to secrets/tokens for security reasons. Please save your secret/token and keep it safe and accessible.",
  "results": {
    "token": "j8s70l647fcdjlmpvedudl4btitm4luh"
  }
}

Keep your tokens accessible in a secure location. For security reasons, TigerGraph Support cannot restore access to tokens.

Parameters

None.

Refresh a token

PUT /requesttoken

A user can refresh a token using either

  • A secret, which is associated with the token to be refreshed

  • Their username and password in their request header

This endpoint takes a token and refreshes the lifetime of the token. The token itself remains unchanged.

Request body

The endpoint expects a JSON request body in the following format:

{
    "token": <token_value>, (1)
    "secret": <secret_value>, (2)
    "lifetime": <token_lifetime> (3)
}
1 Token to be refreshed.
2 User’s secret to generate the token. Required if username and userpassword are not suplied.
3 Period of time for which the token is valid measured in seconds. The default value is about 2.6 million (about a month).

Sample request

  • With secret

  • With username-password pair

  • Response

curl -X PUT http://localhost:9000/requesttoken \
     -d '{"secret": "jiokmfqqfu2f95qs6ug85o89rpkneib3", "token": "tohvf6khjqju8jf0r0l1cohhlm8gi5fq", "lifetime": "15000"}'
curl --user example_username:example_password \
     -X PUT http://localhost:9000/requesttoken \
     -d '{"token": "tohvf6khjqju8jf0r0l1cohhlm8gi5fq", "lifetime": "15000"}'
{
  "code": "REST-0000",
  "expiration": 1641554169,
  "error": false,
  "message": "Refresh token successfully.",
  "token": "0mq98l9pderkaivndf820gudg923p3l0"
}

Parameters

None.

Delete a token

DELETE /requesttoken

A user can delete a token using either

  • A secret, which is associated with the token to be deleted

  • Their username and password in their request header

Request body

The endpoint expects a JSON request body in the following format:

{
    "token": <token_value>, (1)
    "secret": <secret_value>, (2)
}
1 Token to delete.
2 User’s secret to generate the token. Required if username and userpassword are not suplied.

Sample request

  • With secret

  • With username-password pair

  • Response

curl -X DELETE http://localhost:9000/requesttoken \
     -d '{"secret": "jiokmfqqfu2f95qs6ug85o89rpkneib3", "token": "tohvf6khjqju8jf0r0l1cohhlm8gi5fq"}'
curl --user example_username:example_password \
     -X DELETE http://localhost:9000/requesttoken \
     -d '{"token": "tohvf6khjqju8jf0r0l1cohhlm8gi5fq"}'
{
  "code": "REST-0000",
  "expiration": -1,
  "error": false,
  "message": "Drop token successfully.",
  "token": "0mq98l9pderkaivndf820gudg923p3l0"
}

Parameters:

None.

Remove expired tokens

DELETE :8123/expiredtoken

This endpoint removes expired authentication tokens. A user is always allowed to remove their own expired tokens. If a user tries to delete tokens that belong to other users, they need to have the WRITE_USER privilege.

Sample request

The following request deletes all expired tokens. This request requires the privilege WRITE_USER:

curl -X DELETE "https://localhost:8123/expiredtoken"

The following request deletes all expired tokens that belong to users u1 and u2 as well as all tokens created with secrets s1 and s2. This request requires the privilege WRITE_USER

curl -X DELETE 'localhost:8123/expiredtoken' -d '{"user":["u1","u2"],"secret":["s1","s2"]}'

Request body

The request body is optional. If not provided a request body, a request deletes all expired tokens.

{
    "user": ["string"], (1)
    "secret": ["string"] (2)
}
1 Users whose expired tokens to remove. Optional.
2 If a token is created with one of the secrets in the list, they are removed by the request. Optional.

The secrets and users provided in the request body do not have to correlate with each other. A request deletes all tokens associated with any of the users and secrets contained in the payload.

The request is atomic: if any of the provided users or secrets are invalid, or if a user doesn’t have the privilege to remove tokens that belong to one of the users or secrets provided, no tokens will be removed by the request.

Parameters

None.

Loading jobs

Run a Loading Job

POST /ddl/{graph_name}

Required privilege: Graph-level EXECUTE_LOADINGJOB

This endpoint is for loading data into a graph. It submits data as an HTTP request payload, to be loaded into the graph by the DDL Loader. The data payload can be formatted as generic CSV or JSON. For more details, please see GSQL Language Reference Part 1 - Defining Graphs and Loading Data.

If the loading job references multiple files, multiple HTTP requests are needed to complete the loading job since you can only provide data for one filename variable at a time. The loading job will skip the LOAD statements referencing filename variables that the request didn’t provide data for.

To provide data for a filename variable, put the data in the request body and use the filename parameter (explained in the parameter table below) to match the variable name defined in the loading job.

Request body:

The request body is the data to be loaded (either in CSV or JSON format).

Curl allows you to read the data from an input file by using the @ symbol:

curl -X POST --data-binary @./company.csv "http://…"

Sample request:

In this example, the loading job is dependent on three filename variables (f1 and f3) and one filepath string. Therefore, three HTTP requests are needed to complete the loading job.

# Loading job
CREATE LOADING JOB load_data for GRAPH poc_graph {

    DEFINE FILENAME f1;
    DEFINE FILENAME f3;

    LOAD f1 to VERTEX person VALUES ($0, $0);
    LOAD "/home/data/company.csv" to VERTEX company VALUES ($0, $0);

    LOAD f3 to EDGE work_at VALUES ($0, $1, $3, $4, $5);
}

# Provide data for for the second LOAD statement
curl -X POST --data-binary @./another_company.csv "http://localhost:9000/ddl/poc_graph?tag=load_data&filename=__GSQL_FILENAME_0__" | jq

{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "sourceFileName": "Online_POST",
      "statistics": {
        "validLine": 7927,
        "rejectLine": 0,
        "failedConditionLine": 0,
        "notEnoughToken": 0,
        "invalidJson": 0,
        "oversizeToken": 0,
        "vertex": [
          {
            "typeName": "company",
            "validObject": 7,
            "noIdFound": 0,
            "invalidAttribute": 0,
            "invalidPrimaryId": 0,
            "invalidSecondaryId": 0,
            "incorrectFixedBinaryLength": 0
          }
        ],
        "edge": [],
        "deleteVertex": [],
        "deleteEdge": []
      }
    }
  ],
  "code": "REST-0000"
}

# Provide data for filename f1 for the first LOAD statement
curl -X POST --data-binary @./person.csv  "http://localhost:9000/ddl/poc_graph?tag=load_data&filename=f1"

# Provide data for filename f3 for the third LOAD statement
curl -X POST --data-binary @./work_at.csv "http://localhost:9000/ddl/poc_graph?tag=load_data&filename=f3"

Parameters:

Name Required Description

tag

Yes

Loading job name defined in your DDL loading job

filename

Yes

File variable name or file path for the file containing the data

sep

No

Separator of CSV data. If your data is JSON, you do not need to specify this parameter. The default separator is a comma","

eol

No

End-of-line character. Only one or two characters are allowed, except for the special case "\r\n". The default value is "\n"

ack

No

"all": request will return after all GPE instances have acknowledged the POST request. "none": request will return immediately after RESTPP processed the POST request.

timeout

No

Timeout in seconds. If set to 0, use system-wide endpoint timeout setting.

concise

No

Boolean value that indicates whether to return concise results of the data loading request. Concise results will only include the number of vertices and edges added or deleted, and will omit information such as the number of valid and invalid lines in the default response.

If there are special characters in your parameter values, the special characters should use URL encoding. To avoid confusion about whether you should you one or two backslashes, we do not support backslash escapes for the eol or sep parameter.

The maximum size of data you can upload via this endpoint is controlled by the Nginx.ClientMaxBodySize configuration parameter (default is 200 MB).

Graphs

Run built-in functions on graph

POST /builtins/{graph_name}

This endpoint runs a set of built-in functions and returns relevant statistics about a graph.

Request body

This endpoint expects a data payload in the request body that specifies which function to run on the graph. Depending on the function being run, different fields may also be expected in the request body.

Here is a list of functions supported by this endpoint and their corresponding data payload format.

  • stat_vertex_attr

    • Returns the minimum, maximum, and average values of the given vertex type’s int, uint, float and double attributes, and the count of true and false of a boolean attribute.

    • Data payload fields:

      • "function": "stat_vertex_attr": This specifies that the function to run is stat_vertex_attr.

      • "type": The vertex type whose attribute values to report on. Required field. It also accepts the wildcard value *, in which case, all vertex types are included.

  • stat_edge_attr

    • Returns the minimum, maximum, and average values of the given edge type’s int, uint, float and double attributes, and the count of true and false of a boolean attribute.

    • Data payload fields:

      • "function": stat_edge_attr

      • "type": The edge type whose attribute values to report on. Required field. It also accepts the wildcard value * , in which case all edge types are included.

      • "from_type": Optional. The source vertex type of the edges to report on.

      • "to_type": Optional. The target vertex type of the edges to report on.

  • stat_vertex_number

    • Returns the number of vertices of the given vertex type.

    • Data payload fields:

      • "function": "stat_vertex_number"

      • "type": Required field. The vertex type of the vertices to count. It also accepts the wildcard value *, in which case, all vertex types are included.

  • stat_edge_number

    • Returns the number of edges of the given edge type

    • Data payload fields:

      • "function": "stat_edge_number"

      • "type": Required field. The edge type of the edges to count. It also accepts the wildcard value *.

      • "from_type": Optional. The source vertex type of the edges to report on.

      • "to_type": Optional. The target vertex type of the edges to report on.

Sample requests

Below is an example request running stat_vertex_attr on socialNet and its output. The vertex type "Person" has a uint attribute "age".

curl -X POST "http://localhost:9000/builtins/socialNet" \
-d  '{"function":"stat_vertex_attr","type":"Person"}' | jq .

{
  "version": {
      "api": "v2",
      "schema": 0
   },
  "error": false,
  "message": "",
  "results": [
    {
      "vertexName": "Person",
      "attributeStat": [
        {
          "vattrName": "age",
          "MAX": 64,
          "MIN": 15,
          "AVG": 36.5
        }
      ]
    }
  ]
}

Here is an example request running stat_edge_attr on socialNet and its output. The edge type "Liked" has a float attribute "strength".

curl -X POST "http://localhost:9000/builtins/socialNet" \
-d  '{"function":"stat_edge_attr","type":"Liked", "from_type":"*", "to_type":"*"}' | jq .

{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "e_type": "Liked",
      "attributes": {
        "weight": {
          "MAX": 2.5,
          "MIN": 1,
          "AVG": 1.375
        }
      }
    }
  ]
}

Here is an example request running stat_vertex_number and its output.

curl -X POST "http://localhost:9000/builtins/socialNet" \
-d  '{"function":"stat_vertex_number","type":"*"}' | jq .

{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "v_type": "User",
      "count": 4
    },
    {
      "v_type": "Page",
      "count": 4
    },
    {
      "v_type": "Product",
      "count": 7
    },
    {
      "v_type": "DescWord",
      "count": 7
    },
    {
      "v_type": "NameUser",
      "count": 9
    },
    {
      "v_type": "VidUser",
      "count": 4
    },
    {
      "v_type": "Video",
      "count": 5
    },
    {
      "v_type": "AttributeTag",
      "count": 4
    }
  ]
}

Parameters

Name Required Default value Description

realtime

No

False

Force built-in functions to rerun queries instead of using cached data, which is resource-intensive but more accurate if the graph is frequently updated.

Show graph schema metadata

GET /gsqlserver/gsql/schema

Returns schema details about a vertex type, an edge type, or the entire graph schema. This is a GSQL Server request sent to port 14240. Authentication credentials need to be provided.

Sample request:

curl -u tigergraph:tigergraph \
"localhost:14240/gsqlserver/gsql/schema?graph=workNet&type=company"

{
  "error": false,
  "message": "",
  "results": {
    "Config": {
      "STATS": "OUTDEGREE_BY_EDGETYPE",
      "PRIMARY_ID_AS_ATTRIBUTE": false
    },
    "Attributes": [
      {
        "AttributeType": {
          "Name": "STRING"
        },
        "IsPartOfCompositeKey": false,
        "PrimaryIdAsAttribute": false,
        "AttributeName": "id",
        "HasIndex": false,
        "IsPrimaryKey": false
      },
      {
        "AttributeType": {
          "Name": "STRING"
        },
        "IsPartOfCompositeKey": false,
        "PrimaryIdAsAttribute": false,
        "AttributeName": "country",
        "HasIndex": false,
        "IsPrimaryKey": false
      }
    ],
    "PrimaryId": {
      "AttributeType": {
        "Name": "STRING"
      },
      "IsPartOfCompositeKey": false,
      "PrimaryIdAsAttribute": false,
      "AttributeName": "clientId",
      "HasIndex": false,
      "IsPrimaryKey": false
    },
    "Name": "company"
  }
}

Vertex schema object fields:

  • Name: the vertex type name, same as the input parameter "type"

  • PrimaryId: details about the primary id

  • Attributes: details about each attribute, listed in order

  • Config: details about global properties of the vertex type

Edge schema object fields:

  • Name: the edge type name, same as the input parameter "type"

  • FromVertexTypeName: source vertex type name

  • ToVertexTypeName: target vertex type name

  • Attributes: details about each attribute, listed in order

  • IsDirected: whether the edge is directed

  • Config: additional details about global properties of the edge type

Graph schema object fields:

  • GraphName: the graph name, same as the input parameter "graph"

  • VertexTypes: an array of vertex schema objects. Each vertex schema object is exactly the JSON output if that specific vertex type had been specified.

  • EdgeTypes: an array of edge schema objects. Each edge schema object is exactly the JSON output if that specific edge type had been specified.

{
  "error": false,
  "message": "",
  "results": {
    "GraphName": "workNet",
    "VertexTypes": [
      {
        "Config": {...},
        "Attributes": [...],
        "PrimaryId": {...},
        "Name": "person"},
      {
        "Config": {...},
        "Attributes": [...],
        "PrimaryId": {...},
        "Name": "company"}
    ],
    "EdgeTypes": [
      {
        "IsDirected": false,
        "ToVertexTypeName": "company",
        "Config": {},
        "Attributes": [...],
        "FromVertexTypeName": "person",
        "Name": "worksFor"
      }
    ]
  }
}

Parameters:

Name Required Description

graph

Yes

The name of the graph whose schema to retrieve.

type

No

The vertex or edge type whose details to retrieve. If not provided, the endpoint will provide a graph schema object containing the schema details of the entire graph.

Upsert data to graph

The endpoint for upserting data to graph is more complicated than most other REST endpoints and is documented on a separate page. See Upsert data to graph.

Vertices

Insert vertices

To insert vertices or edges, use the Upsert data to graph endpoint.

List vertices

GET /graph/{graph_name}/vertices/{vertex_type}

This endpoint returns all vertices having the type vertex_type in a graph.

This endpoint requires the READ_DATA privilege on the vertex type.

Sample request

  • Request

  • Response

curl -X GET "http://localhost:9000/graph/socialNet/vertices/User"
{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "v_id": "id1",
      "v_type": "User",
      "attributes": {}
    },
    {
      "v_id": "id2",
      "v_type": "User",
      "attributes": {}
    }
    // ... all vertices in graph socialNet of type User
  ]
}

Parameters

Name Required Description

count_only

No

Takes a boolean value. If the value is true, the results field will only contain the count of how many vertices were selected. Default is false.

select

No

Attributes of the selected vertices to return. The parameter takes a list, which is a string of comma-separated values, and will only return the attributes that are provided.

filter

No

Conditions used to filter the returned vertices. The parameter takes a list of conditions, which is a string of comma-separated values. If any filter conditions are provided, the endpoint will only return the vertices that satisfy the conditions. Six comparison operators are supported for this parameter: =, !=, >, >=, < and <=. If the value on the right side of an operator is a string literal, it should be enclosed in double-quotes.

limit

No

Integer value that specifies the total number of vertices to return

sort

No

Attributes to sort the results by. The parameter takes a list, which is a string of comma-separated values, and will sort the returned vertices based on the attributes provided in the list in order. Add "-" in front of the attribute to sort in descending order.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Retrieve a vertex

GET /graph/{graph_name}/vertices/{vertex_type}/{vertex_id}

This endpoint returns a single vertex by its vertex ID. This endpoint requires the READ_DATA privilege on the vertex type.

Sample request

  • Request

  • Response

curl -X GET "http://localhost:9000/graph/socialNet/vertices/User/id1"
{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "v_id": "id1",
      "v_type": "User",
      "attributes": {}
    }
  ]
}

Parameters

Name Required Description

select

No

Attributes of the selected vertices to return. The parameter takes a list, which is a string of comma-separated values, and will only return the attributes that are provided.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Delete vertices

DELETE /graph/{graph_name}/vertices/{vertex_type}

This endpoint deletes vertices by their vertex type. The delete operation is a cascading deletion. If a vertex is deleted, then all edges connected to it are automatically deleted as well.

This endpoint requires the DELETE_DATA privilege on the vertex type.

Sample request

The response object will contain a "deleted_vertices" field that indicates the number of vertices that were deleted.

Request
curl -X DELETE "http://localhost:9000/graph/socialNet/vertices/User"
Response
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": {
    "v_type": "person",
    "deleted_vertices": 3
  }
}

Parameters

Name Required Description

permanent

No

Takes a boolean value. If the value is true, the deleted vertex IDs can never be inserted back, unless the graph is dropped or the graph store is cleared.

filter

No

Conditions used to filter the vertices to delete. The parameter takes a list of conditions, which is a string of comma-separated values. If any filter conditions are provided, the endpoint will only delete the vertices that satisfy the conditions. Six comparison operators are supported for this parameter: =, !=, >, >=, < and . If the value on the right side of an operator is a string literal, it should be enclosed in double-quotes.

limit

No

Integer value that specifies the total number of vertices to delete.

sort

No

Attributes to sort the vertices by. In delete operations,sort should always be used together with limit. The endpoint will delete the number of vertices under the limit specified in the order specified. The parameter takes a list of attributes, and the endpoint will sort all vertices based on the attributes provided in the list in order. Add "-" in front of the attribute to sort by that attribute in descending order.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Delete vertices by type

DELETE /graph/{graph_name}/delete_by_type/vertices/{vertex_type}

This endpoint deletes all vertices of the given vertex type in a graph. This endpoint requires the DELETE_DATA privilege on the vertex type.

Sample request

curl -X DELETE "http://localhost:9000/graph/poc_graph/delete_by_type/vertices/person"

Parameters

Name Required Description

permanent

No

Takes a boolean value. If the value is true, the deleted vertex IDs can never be inserted back, unless the graph is dropped or the graph store is cleared.

ack

No

If the parameter is set to "none", the delete operation doesn’t need to get acknowledgment from any GPE. If it is set to "all" (default), the operation needs to get acknowledgment from all GPEs.

Delete a vertex

DELETE /graph/{graph_name}/vertices/{vertex_type}/{vertex_id}

This endpoint requires the DELETE_DATA privilege on the vertex type.

Sample request

  • Request

  • Response

curl -X DELETE "http://localhost:9000/graph/socialNet/vertices/User/id1"
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": {
    "v_type": "User",
    "deleted_vertices": 1
  }
}

Parameters:

Name Required Description

timeout

no

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Edges

Insert edges

To insert vertices or edges, use the Upsert data to graph endpoint.

List edges of a vertex

GET /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}

This endpoint returns all edges which are connected to a given vertex ID in the graph.

This endpoint requires the READ_DATA privilege on the types or attributes being queried.

Sample request

  • Request

  • Response

curl -X GET "http://localhost:9000/graph/socialNet/edges/VidUser/0?limit=2
{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "2",
      "to_type": "Video",
      "attributes": {
        "rating": 5.2,
        "date_time": 0
      }
    },
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "0",
      "to_type": "Video",
      "attributes": {
        "rating": 6.8,
        "date_time": 0
      }
    }
  ]
}

Parameters:

Name Required Description

count_only

No

Takes a boolean value. If the value is true, the results field will only contain the count of how many edges were selected. Default is false.

select

No

Attributes of the selected edges to return. The parameter takes a list, which is a string of comma-separated values. If select is provided, the edges returned will only show the attributes provided.

filter

No

Conditions used to filter the edges to return. The parameter takes a list of conditions, which is a string of comma-separated values. If any filter conditions are provided, the endpoint will only return the edges that satisfy the conditions. Six comparison operators are supported for this parameter: =, !=, >, >=, < and <=. If the value on the right side of an operator is a string literal, it should be enclosed in double-quotes.

limit

No

Integer value that specifies the maximum limit of the total number of edges to return.

sort

No

Attributes to sort the results by. The parameter takes a list, which is a string of comma-separated values, and will sort all the edges based on the attributes provided in the list in order. Add "-" in front of the attribute to sort in descending order.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, system-wide endpoint timeout setting is applied.

List edges of a vertex by edge type

GET /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}/{edge_type}

This endpoint lists all the edges of a specified type connected to a given vertex ID in the graph.

This endpoint requires the READ_DATA privilege on the types or attributes being queried.

Sample request

curl -X GET "http://localhost:9000/graph/socialNet/edges/VidUser/0/User_Video?limit=2

{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "2",
      "to_type": "Video",
      "attributes": {
        "rating": 5.2,
        "date_time": 0
      }
    },
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "0",
      "to_type": "Video",
      "attributes": {
        "rating": 6.8,
        "date_time": 0
      }
    }
  ]
}

Parameters:

Name Required Description

count_only

No

Takes a boolean value. If the value is true, the results field will only contain the count of how many edges were selected. Default is false.

select

No

Attributes of the selected edges to return. The parameter takes a list, which is a string of comma-separated values. If select is provided, the edges returned will only show the attributes provided.

filter

No

Conditions used to filter the edges to return. The parameter takes a list of conditions, which is a string of comma-separated values. If any filter conditions are provided, the endpoint will only return the edges that satisfy the conditions. Six comparison operators are supported for this parameter: =, !=, >, >=, < and <=. If the value on the right side of an operator is a string literal, it should be enclosed in double quotes.

limit

No

Integer value that specifies the maximum limit of the total number of edges to return.

sort

No

Attributes to sort the results by. The parameter takes a list, which is a string of comma-separated values, and will sort all the edges based on the attributes provided in the list in order. Add "-" in front of the attribute to sort in descending order.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, system-wide endpoint timeout setting is applied.

List edges of a vertex by edge type and target type

GET /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}/{edge_type}/{target_vertex_type}

This endpoint lists edges connected to a given vertex by edge type and target vertex type.

This endpoint requires the READ_DATA privilege on the types or attributes being queried.

Use "_" for edge_type in the URL to permit any edge type.

Sample request

  • Request

  • Response

curl -X GET "http://localhost:9000/graph/socialNet/edges/VidUser/0/User_Video/Video?limit=2
{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "2",
      "to_type": "Video",
      "attributes": {
        "rating": 5.2,
        "date_time": 0
      }
    },
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "0",
      "to_type": "Video",
      "attributes": {
        "rating": 6.8,
        "date_time": 0
      }
    }
  ]
}

Parameters

Name Required Description

count_only

No

Takes a boolean value. If the value is true, the results field will only contain the count of how many edges were selected. Default is false.

not_wildcard

No

Boolean value that indicates whether or not "_" supplied in the endpoint URL is a wildcard. If the parameter is true, "_" is interpreted literally to select only edges with edge type name equal to underscore.

select

No

Attributes of the selected edges to return. The parameter takes a list, which is a string of comma-separated values. If select is provided, the edges returned will only show the attributes provided.

filter

No

Conditions used to filter the edges to return. The parameter takes a list of conditions, which is a string of comma-separated values. If any filter conditions are provided, the endpoint will only return the edges that satisfy the conditions. Six comparison operators are supported for this parameter: =, !=, >, >=, < and <=. If the value on the right side of an operator is a string literal, it should be enclosed in double-quotes.

limit

No

Integer value that specifies the maximum limit of the total number of edges to return.

sort

No

Attributes to sort the results by. The parameter takes a list, which is a string of comma-separated values, and will sort all the edges based on the attributes provided in the list in order. Add "-" in front of the attribute to sort in descending order.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Retrieve edge by source, target, and edge type

GET /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}/{edge_type}/{target_vertex_type}/{target_vertex_id}

This endpoint returns the edge of a specified type between a source vertex and a target vertex. If the edge type isn’t defined with a discriminator, the source, target and edge type uniquely identify an edge. If the edge type is defined with a discriminator, this endpoint returns all edges of the edge type between the source and target vertices.

This endpoint requires the READ_DATA privilege on the types or attributes being queried.

Sample request

  • Request

  • Response

curl -X GET "http://localhost:9000/graph/socialNet/edges/VidUser/0/User_Video/Video/2"
{
  "version": {
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "e_type": "User_Video",
      "directed": false,
      "from_id": "0",
      "from_type": "VidUser",
      "to_id": "2",
      "to_type": "Video",
      "attributes": {
        "rating": 5.2,
        "date_time": 0
      }
    }
   ]
 }

Parameters

Name Required Description

select

No

Attributes of the selected edges to return. The parameter takes a list, which is a string of comma-separated values. If select is provided, the edges returned will only show the attributes provided.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Retrieve edge by source, target, edge type, and discriminator

GET /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}/{edge_type}
/{target_vertex_type}/{target_vertex_id}/{discriminator}

This endpoint allows you to retrieve an edge by its source, target, edge type, and discriminator.

If the discriminator has multiple attributes, use , as a separator. But if the attributes already have commas in a string discriminator, use \ as an escape character to escape the commas. The backslash escape character can escape the escape character itself as well.

This endpoint requires the READ_DATA privilege on the types or attributes being queried.

Sample request

  • Single-attribute discriminator

  • Composite discriminator

curl -X GET 'http://localhost:9000/graph/multiedge/edges/Person/4/transfer/Person/30/2431'
curl -X GET "http://localhost:9000/graph/multiedge/edges/Person/9/transfer2/Account/9/77,1117,2010-04-09%2003:57:01" (1)
1 %20 is the URL encoding for a space character.

Parameters

Name Required Description

select

No

Attributes of the selected edges to return. The parameter takes a list, which is a string of comma-separated values. If select is provided, the edges returned will only show the attributes provided.

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Delete an edge by source, target, and edge type

DELETE /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}/{edge_type}/{target_vertex_type}/{target_vertex_id}

Deletes an edge by its source vertex type and ID, target vertex type and ID, as well as edge type. If the edge type isn’t defined with a discriminator, the source, target and edge type uniquely identify an edge.

If the edge type is defined with a discriminator, this endpoint deletes all edges of the edge type between the source and target vertices. If you want to delete a specific edge by its discriminator, see Delete an edge by source, target, edge type and discriminator.

This endpoint requires the DELETE_DATA privilege and READ_DATA on the types or attributes being queried.

Sample request

  • Request

  • Response

curl -X DELETE "https://crunch.i.tgcloud.io:9000/graph/CrunchBasePre_2013/edges/person/p:23601/work_for_company/company/c:14478"
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "e_type": "work_for_company",
      "deleted_edges": 1
    }
  ]
}

Parameters

Name Required Description

timeout

No

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Delete an edge by source, target, edge type and discriminator

DELETE /graph/{graph_name}/edges/{source_vertex_type}/{source_vertex_id}/{edge_type}
/{target_vertex_type}/{target_vertex_id}/{discriminator}

This endpoint allows you to delete an edge by its source, target, edge type, and discriminator.

If the discriminator has multiple attributes, use , as a separator. But if the attributes already have commas in a string discriminator, use \ as an escape character to escape the commas. The backslash escape character can escape the escape character itself as well.

This endpoint requires the DELETE_DATA privilege on the type of the edge being queried.

Sample request

  • Single-attribute discriminator

  • Composite discriminator

curl -X DELETE 'http://localhost:9000/graph/multiedge/edges/Person/4/transfer/Person/30/2431'
curl -X DELETE "http://localhost:9000/graph/multiedge/edges/Person/9/transfer2/Account/9/77,1117,2010-04-09%2003:57:01" (1)
1 %20 is the URL encoding for a space character.

Parameters

Name Required Description

timeout

no

Integer that specifies the number of seconds after which the query will time out. If the parameter is set to 0 or isn’t provided, the system-wide endpoint timeout setting is applied.

Queries

Get query metadata

GET /gsqlserver/gsql/queryinfo

Returns metadata details about a query. In particular, it lists the input parameters in the same order as they exist in the query (in the form of a JSON array of JSON objects) and outputs PRINT statement syntax. This endpoint exists on port 14240 and requests are sent to the GSQL server. Therefore, you should provide authentication credentials in the request.

Sample request

  • Request

  • Response

curl -u tigergraph:tigergraph -X GET "http://localhost:14240/gsqlserver/gsql/queryinfo?graph=workNet&query=to_vertex_setTest"
{
  "output": [
    {
      "v": "vertex"
    },
    {
      "@@v2": "SetAccum<vertex>"
    },
    {
      "S2": [
        {
          "v_id": "int",
          "attributes": {
            "interestList": "INT_LIST",
            "skillSet": "INT_SET",
            "skillList": "INT_LIST",
            "locationId": "STRING",
            "interestSet": "INT_SET",
            "id": "STRING"
          },
          "v_type": "person"
        },
        {
          "v_id": "int",
          "attributes": {
            "country": "STRING",
            "id": "STRING"
          },
          "v_type": "company"
        }
      ]
    },
    {
      "SDIFF.size()": "int"
    }
  ],
  "input": [
    {
      "uid": "string"
    },
    {
      "uids": "set<string>"
    },
    {
      "vtype": "string"
    }
  ],
  "queryname": "to_vertex_setTest",
  "error": false,
  "message": "",
  "version": {
    "schema": 0,
    "edition": "DEVELOPER_EDITION",
    "api": "V2"
  }
}

The JSON response object contains three fields:

  • queryname: name of the query, same as the query input parameter.

  • input: Ordered list of the input parameter names and data types in the same order as they are in the query.

  • output: JSON object that follows the same structure of the query’s output. For each key-value pair, the key is the name that appears in the query output, while the values are the data types of the output.

Parameters:

Name Required Description

graph

Yes

Name of the graph

query

Yes

Name of the query

Install a query

POST /gsql/queries/install

Once queries has been created for a graph, the user can use the POST /gsql/queries/install endpoint to make a request to install one or multiple queries.

Once the GSQL server has received the POST request, it creates an installation request and returns a request location containing a request ID, with which you can check the installation status of all the queries in the request. An installation request will wait in the background for any currently running processes to complete.

Parameters

Name Required Description

graph

Yes

The name of the graph the queries belong to

queries

Yes

The names of the queries to install. If queries is set to all or *, the endpoint installs all queries.

To pass in the names of multiple queries, enter the same parameter multiple times with the names of the queries you want to install.

Sample request

In the below example, the user requests to install all queries on the graph poc_graph by using the * wildcard.

  • Request

  • Response

curl --user tigergraph:tigergraph -X POST 'http://localhost:8123/gsql/queries/install?graph=poc_graph&queries=*'
{
    "location": "/gsql/queries/install/1234", (1)
    "error": false,
    "message": "Request 1234 successfully created",
    "results": ""
}
1 URI where the request can be found. 1234 is the ID of the request.

Check query installation status

GET /gsql/queries/install/{request_id}

This endpoint takes a request location generated by the POST /gsql/queries/install endpoint and returns the state of the installation request for each query, as well as the overall state of the request itself.

Parameters

None.

Sample request

In the below example, the user requests information about the queries installed in the above POST example. The response shows that three query install requests were made with POST: queries t1, t2, and t3.

At the time of response, t1 had installed successfully, while the other two were still running in the background.

  • Request

  • Response

curl --user tigergraph:tigergraph -X GET 'http://localhost:8123/gsql/queries/install/1234'
{
  "state": "RUNNING", (1)
  "error": false,
  "message": "",
  "results": "",
  "queries": [
    {
      "name": "t1",
      "state": "SUCCESS"
    },
    {
      "name": "t2",
      "state": "RUNNING"
    },
    {
      "name": "t3",
      "state": "RUNNING"
    }
  ],
  "graph": "poc_graph"
}
1 The installation request state shown with state shows the overall status of the query installation request. It will return SUCCESS even if some queries fail to install, as long as each query has been given a chance to install. If the overall status is SUCCESS but each query has failed, it indicates that something is wrong with the queries. If the overall status is FAILURE, it indicates that something is wrong with the installation job itself independent of the queries. It will return NEW if the query installation request has been made, but the resource manager is occupied with another request at the time the user made the overall status request.

If you install all queries and receive a state of SUCCESS for one or more queries that had been installed previously, it does not mean that they were reinstalled. A returned state of SUCCESS is a confirmation that a query is ready to be run.

Run an installed query (GET)

GET /query/{graph_name}/{query_name}

Each time a new TigerGraph query is installed, a dynamic endpoint is generated. This new endpoint enables the user to run the new TigerGraph query through HTTP requests and giving the parameters in URL or in a data payload. In the case of a GET request, parameters should be passed in through the query string.

Parameters

Name Required Description

read_committed

No

Boolean value that indicates whether to use read-committed isolation level for the query. At the read committed level, it is guaranteed that any data read is committed at the moment it is read. By default, it is off.

To use the read_commited parameter, read_commited must be defined as a parameter in the query itself. The content of the query does not have to use the parameter, but the parameter must be defined. For example:

CREATE QUERY PageRank( …​, INT iteration, …​, BOOL read_committed) FOR GRAPH Example_Graph

Query parameter passing

When using a GET request to run an installed query, the query parameters are passed in through the query string of the URL.

Parameter type Query string format Example

Set or bag of primitives

Assign multiple values to the same parameter name.

A set p1 of integers: p1=1&p1=5&p1=10

VERTEX<type>

Use the ID of the vertex: parameterName=vertex_id

A vertex with parameter name vp and an ID of person2: vp=person2

VERTEX (type not pre-specified)

Use two query string parameters:

  • parameterName=vertex_id

  • parameterName.type=vertex_type

A vertex with parameter name va , type person and ID person1:

va=person1&va.type=person

Set or bag of VERTEX<type>

Assign multiple vertex IDs to the same SET or BAG parameter name.

A set parameter named vp of vertices of type person:

vp=person3&vp=person4

Set or bag of VERTEX (type not pre-specified)

The SET or BAG must be treated as an array. A vertex ID and type must be provided for the vertex element at each index.

A set parameter named vp of vertices:

vp[0]=person1&vp[0].type=person&vp[1]=11&vp[1].type=post

Specify replica

If you have a TigerGraph HA cluster, you can specify a query to run on a particular replica with the HTTP header GSQL-REPLICA. The value of the header needs to be an integer within the range [1, (cluster replication factor)]. If you supply an invalid value for the header, the request will return an error.

Specify thread limit

When running a query through RESTPP, you can specify a limit on the number of threads that the query is allowed to use on each node through the HTTP header GSQL-THREAD-LIMIT. The number of threads used by a query means the number of vCPUs used by the query. By default, a query will use all threads that are available on a machine.

For example, if you have a cluster of three nodes, each with 8 vCPUs, then a query will use all 8 threads available on a node in the cluster by default. By providing a thread limit in the request header, you can limit the query to only use a number of threads under the limit.

Sample requests

To run a query that takes a parameter of a set of strings (str), a vertex parameter(v), and a parameter of a set of vertices:

  • Query

  • Request

CREATE QUERY print_params(SET<STRING> str, VERTEX v, SET<VERTEX> party)
FOR GRAPH Social {
  PRINT str, v, party;
}

To pass in the value for a set parameter, enter the values of all set elements to the name of the parameter multiple times. If the type of a vertex is not specified, specify it by adding a parameter <typeName>.type and specify the type:

$ curl -X GET "http://localhost:9000/query/social/print_params?str=hello&str=world&v=Mary&party=Tobias&party=Jenny&party.type=person&v.type=person"

To run query hello on a graph named social, and the query parameter is of type VERTEX<person> whose ID is "Tom"

Running a query via HTTP request
curl -X GET "http://localhost:9000/query/social/hello?p=Tom"
Limiting the query to use under 4 threads
curl -x GET -H "GSQL-THREAD-LIMIT: 4" "http://localhost:9000/query/social/hello?p=Tom"
Specifying the query to run on the first replica
curl -X GET -H "GSQL-REPLICA: 2" "http://localhost:9000/query/social/hello?p=Tom"
Specifying the query to run on the primary cluster
curl -X GET -H "GSQL-REPLICA: 1" "http://localhost:9000/query/social/hello?p=Tom"

Run an installed query (POST)

POST /query/{graph_name}/{query_name}

Users can also run queries through a POST request, which allows them to pass query parameters in JSON format. This is especially helpful when the query takes complex parameters.

Format of file in JSON format

{
    "parameter_1":"value_1",
    "parameter_2":"value_2"
}

Format of POST request

curl -X POST --data-binary @./<sample_file> "http://localhost:9000/query/<sample_graph>/<sample_query>"

Parameters

Name Required Description

read_committed

No

Boolean value that indicates whether to use read-committed isolation level for the query. At the read committed level, it is guaranteed that any data read is committed at the moment it is read. By default, it is off.

Query parameter Passing

When using a POST request to run an installed query, the query parameters are passed in through the request body and encoded in JSON format. The formatting rules for the JSON payload are the same as using JSON to pass in parameters in the RUN QUERY command.

Parameter type Syntax Example

DATETIME

Use a string formatted as "YYYY-MM-DD HH-MM-SS"

"2019-02-19 19:19:19"

Set or bag of primitives

Use a JSON array containing the primitive values

["a", "list", "of", "args"]

VERTEX<type>

Use a JSON object containing a field "id" for the vertex ID and a field "type" for the type of the vertex

{"id": "person1", "type": "person"}

VERTEX (type not specified)

Use a JSON object containing a field "id" for the vertex ID

{"id": "person1"}

Set or bag of VERTEX<type>

Use a JSON array containing a list of JSON VERTEX<type> object

[{"id": "person1"}, {"id": "person2"}]

Set or bag of vertices of unspecified types

Use a JSON array containing a list of JSON VERTEX

[{"id": "person1","type": "person"},{"id": "person2","type": "person"}]

Headers

The Run Query endpoint accepts a number of headers that allow you to configure the way the query is run:

Name Description

GSQL-REPLICA

The replica where the query is to be run. The value of the header needs to be an integer within the range one to the replication factor of the cluster. If you supply an invalid value for the header, the request returns an error.

A value of 0 specifies that the query be run on the primary copy; a value of 1 specifies that the query be run on the first replica, and so on.

GSQL-THREAD-LIMIT

Maximum limit on the number of threads that the query is allowed to use on each node. By default, a query will use all threads that are available on a machine.

For example, if you have a cluster of three nodes, each with 8 vCPUs, then a query will use all 8 threads available on a node in the cluster by default. By providing a thread limit in the request header, you can limit the query to only use a number of threads under the limit.

GSQL-QueryLocalMemLimitMB

Maximum limit of how much memory a query is allowed to use on any single node in a cluster. By default, there is no limit to the amount of memory a query is allowed to consume until the overall free memory reaches a critical level.

Sample request

To run a query that takes a parameter of a set of strings (str), a vertex parameter(v), and a parameter of a set of vertices (party):

  • Query

  • Request

  • Request body

CREATE QUERY print_params(SET<STRING> str, VERTEX v, SET<VERTEX<Person>> party) FOR GRAPH Social {
  PRINT str, v, party;
}
$ curl -X POST -d @<path_to_request_body> ""http://localhost:9000/query/Social/print_params"
{
	"v": {"id": "Jenny", "type": "Person"}, (1)
	"str": ["hello", "world"], (2)
	"party": [
		{"id": "Mary"},
		{"id": "Tobias}
	] (3)
}
1 Provides the value for parameter v, which is a Person type vertex with ID Jenny. The type field is required because the original query definition does not specify vertex type.
2 Provides the value for parameter str, which is a set of strings with two elements "hello" and "world".
3 Provides the value for parameter party, which is a set of Person type vertices. The type field is optional for the vertex elements in the set because the original query definition specifies vertex type.

The query in this sample request takes a parameter of type VERTEX<person>:

curl -X POST -d '{"p":{"id":"Tom","type":"person"}}'
"http://localhost:9000/query/social/hello"
Specify that the query run on the first replica
curl -X POST -H "GSQL-REPLICA: 2" -d '{"p":{"id":"Tom","type":"person"}}'
"http://localhost:9000/query/social/hello"
Specify that the query run on the primary cluster
curl -X POST -H "GSQL-REPLICA: 1" -d '{"p":{"id":"Tom","type":"person"}}'
"http://localhost:9000/query/social/hello"
Specify that the query run with a limit of 4 threads
curl -X POST -H "GSQL-THREAD-LIMIT: 4" -d '{"p":{"id":"Tom","type":"person"}}' "http://localhost:9000/query/social/hello"

Installed queries can run in Detached Mode. To do this, use the GSQL-ASYNCheader and set its value to true.

The results and status of the queries run in Detached Mode can be retrieved with a query ID, which is returned immediately when queries are executed in Detached Mode.

Run an interpreted query

POST /gsqlserver/interpreted_query

This endpoint runs a GSQL query in Interpreted Mode. The query body should be supplied at the data payload, and the query’s parameters are supplied as the URL’s query string. This endpoint exists on the GSQL server on port 14240.

This request goes directly to the GSQL server (port 14240) instead of the RESTPP server (port 9000), so the username and password must be specified in the header. If you are using curl, you can use the -u option as shown below.

Request body:

The request body for this endpoint should be the entire INTERPRET QUERY statement.

Parameter passing:

When running an interpreted query through this endpoint, the query parameters should be passed in through the URL query string.

Sample request:

curl --fail -u <my_username>:<my_password> -X POST "http://localhost:14240/gsqlserver/interpreted_query?a=10" -d 'INTERPRET QUERY (INT a) FOR GRAPH gsql_demo {
    PRINT a;
 }'

List running queries of a graph

GET /showprocesslist/{graph_name}

This endpoint reports statistics of running queries of a graph: the query’s request ID, start time, expiration time, and the REST endpoint’s URL. This includes installed queries. interpreted queries, as well as built-in queries.

If you are running a TigerGraph cluster, this endpoint only shows queries running on the node to which the request is sent, not all running queries on the cluster.

This endpoint requires the READ_DATA privilege on the graph you are querying.

Sample request

  • Request

  • Response

curl -X GET "http://localhost:9000/showprocesslist/poc_graph" | jq .
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "requestid": "65538.RESTPP_1_1.1558386411523.N",
      "startTime": "2019-05-20 14:06:51.523",
      "expirationTime": "2019-05-20 14:15:11.523",
      "url": "/sleepgpe?milliseconds=100001"
    },
    {
      "requestid": "196609.RESTPP_1_1.1558386401478.N",
      "startTime": "2019-05-20 14:06:41.478",
      "expirationTime": "2019-05-20 14:15:01.478",
      "url": "/sleepgpe?milliseconds=100000"
    }
  ],
  "code": "REST-0000"
}

Parameters

No Parameters.

List all running queries

GET /showprocesslistall

This endpoint reports the statistics of all running queries across all graphs on a TigerGraph instance: the query’s request ID, start time, expiration time, and the REST endpoint’s URL. This includes installed queries. interpreted queries, as well as built-in queries.

If you are running a TigerGraph cluster, this endpoint only shows queries running on the node to which the request is sent, not all running queries on the cluster.

If a high volume of queries are running, invoking the endpoint can be an expensive operation that should be used with caution.

This endpoint requires the READ_DATA privilege on the global scope.

Sample request

  • Request

  • Response

$ curl -X GET "http://localhost:9000/showprocesslistall" |jq
  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100   671  100   671    0     0   655k      0 --:--:-- --:--:-- --:--:--  655k
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "requestid": "15.RESTPP_1_1.1660165648633.N", (1)
      "startTime": "2022-08-10 21:07:28.633",
      "expirationTime": "2022-08-10 21:07:44.633", (2)
      "url": "/query/Social/rngExample",
      "elapsedTime": 3385
    },
    {
      "requestid": "65551.RESTPP_1_1.1660165648638.N",
      "startTime": "2022-08-10 21:07:28.638",
      "expirationTime": "2022-08-10 21:07:44.638",
      "url": "/query/Work_Net/rngExample",
      "elapsedTime": 3380
    },
    {
      "requestid": "131087.RESTPP_1_1.1660165648723.N",
      "startTime": "2022-08-10 21:07:28.723",
      "expirationTime": "2022-08-10 21:07:44.723",
      "url": "/query/Entity_Resolution/rngExample",
      "elapsedTime": 3295
    }
  ],
  "code": "REST-0000"
}
1 Each request to run a query sent to the REST server has a unique ID that can be used to look up the request in the system logs.
2 The time at which the query will time out.

Parameters

No parameters.

Abort a query

GET /abortquery/{graph_name}

Required privilege: Graph-level DELETE_DATA

This endpoint safely aborts a selected query by ID or all queries of an endpoint by endpoint URL of a graph.

If you are running a TigerGraph cluster, this endpoint only allows you to abort a query sent to the requested node, not any query in the cluster.

Sample request

  • Request

  • Response

curl -X GET "localhost:9000/abortquery/poc_graph?requestid=16842763.RESTPP_1_1.1561401340785.N&requestid=16973833.RESTPP_1_1.1561401288421.N"
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [
    {
      "aborted_queries": [
        {
          "requestid": "16842763.RESTPP_1_1.1561401340785.N",
          "url": "/sleepgpe?milliseconds=110000"
        },
        {
          "requestid": "16973833.RESTPP_1_1.1561401288421.N",
          "url": "/sleepgpe?milliseconds=100000"
        }
      ]
    }
  ],
  "code": "REST-0000"
}

Parameters

Name Required Description

requestid

No

The ID of the query to abort. It can take a single query ID or the string "all". If requestid is set to all. It will abort all running queries.

url

No

The endpoint whose running queries to abort. You must specify the base of the endpoint’s URL, but then use a wildcard to allow for different parameters. For example, to abort all running queries for the endpoint /sleepgpe, use url=/sleepgpe.*

Check query status (Detached Mode)

GET /query_status

This endpoint allows you to check the status of a query run in detached mode.

If you are running a TigerGraph cluster, this endpoint only allows you to check the status of a query running on the node to which the request is sent, not all nodes on the cluster.

Parameters

Parameter Required Description

requestid

No

String ID of the request. The parameter can have the value all, in which case the endpoint returns one status report for each running request. If not specified, the value of the parameter defaults to all.

graph_name

No

The name of the graph whose running queries to report on. If not specified, the endpoint returns queries running on the first created graph on the node to which the request is sent.

Sample request

  • Request

  • Response

curl -s -X GET "http://localhost:9000/query_status?graph_name=poc_graph&requestid=4.RESTPP_1_1.1599672031541.N"
{
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "error": false,
  "message": "",
  "results": [{
    "requestid": "4.RESTPP_1_1.1599672031541.N", (1)
    "startTime": "2020-09-09 10:20:31.541", (2)
    "expirationTime": "2020-09-09 10:20:47.541", (3)
    "url": "/query/ldbc_snb/countIndirectFriends?pid=21990232555889", (4)
    "elapsedTime": 19, (5)
    "status": "success" (6)
  }]
}
1 The query ID associated with the given query status JSON object.
2 The timestamp for the start time of the given query.
3 The timestamp for when the given query times out. The default timeout limit is 16 seconds and can be set using the GSQL-TIMEOUT header.
4 URL of the given query.
5 Elapsed real time of the given query measured in milliseconds. For completed queries, the value shows the total runtime of the request. For ongoing queries, it shows the amount of time taken so far.
6 The status of the given query. Possible values are “success”, “timeout”, “aborted”, or “running”.

If one or more of the provided query IDs (requestid) are invalid, the return JSON will include an unknown_requestid field containing all the invalid query IDs. If a query ID is marked as unknown, it means either the query does not exist or that it was not run in Detached Mode.

Check query results (Detached Mode)

GET /query_result

This endpoint allows you to check the results of queries run in Detached Mode if they have finished running. If the query is still running, the endpoint will respond with an error and a message saying "Unable to retrieve result for query <requestid>". Ensure that the query is finished before checking its result.

This endpoint only allows you to check the results of a query sent to the requested node, not any query in the graph.

Sample request

  • Request

  • Response

curl -s -X GET "http://localhost:9000/query_result?requestid=196611.RESTPP_1_1.1630601692834.N"
{
  "error": false,
  "message": "",
  "version": {
    "edition": "enterprise",
    "api": "v2",
    "schema": 0
  },
  "results": [{"vSet": [{
    "v_id": "21990232555889",
    "attributes": {"vSet.@friendCount": 13677},
    "v_type": "Person"
  }]}]
}

Parameters

Name Required Description

requestid

Yes

String ID of the query.

Path-Finding Algorithms

The TigerGraph platform comes with two built-in endpoints, /shortestpath and /allpaths, which return either the shortest or all unweighted paths connecting a set of source vertices to a set of target vertices. The table below summarizes the two path-finding endpoints.

Input Parameters and Output Format for Path-Finding

Each REST endpoint reads a JSON-formatted payload that describes the input parameters. These parameters specify which vertices and edges may be on the paths, additional conditions on the attributes of the vertices and edges, and the maximum length of a path.

Source and target vertices

Each endpoint must have either a source or sources key and either a target or targets parameter. The source and target parameters describe a single vertex. The format for a vertex object is as follows:

{"type" : "<vertex_type_name>", "id" : "<vertex_id>"}.

The sources and targets parameters are JSON arrays containing a list of vertex objects.

Filters
The payload may also have an array of filter conditions, to restrict the vertices or edges in the paths. Each individual filter is a JSON object which describes a condition on one vertex type or edge type. A filter object has one or two key-value pairs:

{"type": "<vertex_or_edge_type>", "condition": "<attribute_condition>"}

  • "type": the vertex type or edge type to be filtered

  • "condition" (optional): a boolean expression on one attribute of the given vertex type or edge type. and and or may be used to make compound expressions.

Example of a filter array:

[{"type": "bought", "condition": "price < \"100\" and quality == \"good\""},
 {"type": "sold",   "condition": "price > \"100\"  or quality != \"good\""}]

Note that all filtering conditions in vertexFilters and edgeFilters are combined with the or relationship, i.e., if a vertex (or edge) fulfills any one of the filter conditions, then this vertex (or edge) will be included in the resulting paths.

Output

The JSON output is a list of vertices and a list of edges. Each vertex and each edge is listed in full with all attributes. The collections of vertices and edges are not in path order.

Find shortest path

POST /shortestpath/{graph_name}

This endpoint takes a source vertex or a set of source vertices and a target vertex or a set of target vertices. It returns the shortest path between the source and the target. If the source is a set of vertices, the resulting path will begin with one of the vertices in the set. If the target is a set of vertices, the resulting path will end with one of the vertices in the set.

Request body:

This endpoint expects a request body that describes the source and target vertex or vertex set. Below is a table of all the fields in the request body.

Key Type Description

source

vertex object

Each path must start from this vertex. Mutually exclusive with sources.

sources

vertex array

Each path must start from one of these vertices. Mutually exclusive with source.

target

vertex object

Each path must end at this vertex. Mutually exclusive with targets.

targets

vertex array

Each path must end at one of these vertices. Mutually exclusive with target.

vertexFilters

filter array

(OPTIONAL) Restrict the paths to those whose vertices satisfy any of the given filters.

edgeFilters

filter array

(OPTIONAL) Restrict the paths to those whose edges satisfy any of the given filters. See details of filters above.

Sample request:

  • Request

  • Response

curl -s -X POST "http://localhost:9000/shortestpath/movieNet" \
-d '{
  "sources":[{"type":"VidUser","id":"2"}],
  "targets":[{"type":"VidUser","id":"0"}, {"type":"VidUser","id":"3"}],
  "edgeFilters":[{"type":"User_Video","condition":"rating > 5 and date_time > 1000"}],
  "maxLength":4
}'
{
  "version": { "edition": "developer", "api": "v2", "schema": 0 },
  "error": false,
  "message": "Cannot get 'vertex_filters' filters, use empty filter.",
  "results": [
    {
      "vertices": [
        { "v_id": "3","v_type": "VidUser","attributes": { "name": "Dale" }},
        { "v_id": "0","v_type": "Video","attributes": { "name": "Solo", "year", 2018 }},
        { "v_id": "0","v_type": "VidUser","attributes": { "name": "Angel" }},
      ],
      "edges": [
        {
          "e_type": "User_Video", "from_id": "0", "from_type": "Video",
          "to_id": "0", "to_type": "VidUser", "directed": false,
          "attributes": { "rating": 6.8, "date_time": 15000 }
        },
        {
          "e_type": "User_Video", "from_id": "0", "from_type": "Video",
          "to_id": "3", "to_type": "VidUser",  "directed": false,
          "attributes": { "rating": 6.6, "date_time": 16000 }
        }
      ]
    }
  ]
}

Parameters:

Key Required Description

maxLength

No

Integer that specified the maximum length of a shortest path. The default value is 6.

allShortestPaths

No

If true, the endpoint will return all shortest paths between the source and target. Default is false, meaning that the endpoint will return only one path.

Find all paths

POST /allpaths/{graph_name}

This endpoint finds all paths between a source vertex (or vertex set) and target vertex (or vertex set).

Request body:

This endpoint expects a request body that describes the source and target vertex or vertex set. Below is a table of all the fields in the request body.

Key Type Description

source

vertex object

Each path must start from this vertex. Mutually exclusive with sources.

sources

vertex array

Each path must start from one of these vertices. Mutually exclusive with source.

target

vertex object

Each path must end at this vertex. Mutually exclusive with targets.

targets

vertex array

Each path must end at one of these vertices. Mutually exclusive with target.

vertexFilters

filter array

(OPTIONAL) Restrict the paths to those whose vertices satisfy any of the given filters.

edgeFilters

filter array

(OPTIONAL) Restrict the paths to those whose edges satisfy any of the given filters. See details of filters above.

Parameters:

Name Required Description

maxLength

Yes

Maximum path length.

The current implementation of this endpoint will include paths with loops. Since it is possible to go around a loop an infinite number of times, it is important that you select the smallest value of maxLength which you consider appropriate. Even if there are no loops in your graph, a smaller maxLength will allow your query to run faster.

Sample request:

The example below requests all paths up to maximum length 3 between the source vertex set {Video 0} and the target vertex set {AttributeTag "action"}. The path may only contain Video vertices where year >= 1984. The result includes 3 paths:
AttrributeTag "action"  —  Video 0
AttrributeTag "action"  —  Video 3 — VidUser 4 — Video 0
AttrributeTag "action"  —  Video 2 — VidUser 0 — Video 0

  • Request

  • Response

curl -s -X POST "http://localhost:9000/allpaths/movieNet" -d '{
  "sources":[{"type":"Video","id":"0"}],
  "targets":[{"type": "AttributeTag", "id":"action"}],
  "vertexFilters":[{"type":"Video", "condition":"year >= 1984"}],
  "maxLength": 3
}'
{
  "version": { "edition": "developer", "api": "v2", "schema": 0 },
  "error": false,
  "message": "Cannot get 'edge_filters' filters, use empty filter.",
  "results": [
    {
      "vertices": [
        { "v_id": "action","v_type": "AttributeTag","attributes": {}},
        { "v_id": "3","v_type": "VidUser","attributes": { "name": "Dale" }},
        { "v_id": "0","v_type": "VidUser","attributes": { "name": "Angel" }},
        { "v_id": "0","v_type": "Video","attributes": { "name": "Solo", "year": 2018 }},
        { "v_id": "2","v_type": "Video","attributes": { "name": "Thor", "year": 2011 }},
        { "v_id": "4","v_type": "Video","attributes": { "name": "Ran", "year": 1985 }}
      ],
      "edges": [
        {
          "e_type": "Video_AttributeTag", "from_id": "0", "from_type": "Video",
          "to_id": "action", "to_type": "AttributeTag", "directed": false,
          "attributes": { "weight": 1, "date_time": 0 }
        },
        {
          "e_type": "Video_AttributeTag", "from_id": "4", "from_type": "Video",
          "to_id": "action", "to_type": "AttributeTag", "directed": false,
          "attributes": { "weight": 1, "date_time": 11000 }
        },
        {
          "e_type": "User_Video", "from_id": "3", "from_type": "VidUser",
          "to_id": "4", "to_type": "Video", "directed": false,
          "attributes": { "rating": 8.4, "date_time": 12000 }
        },
        {
          "e_type": "User_Video", "from_id": "3", "from_type": "VidUser",
          "to_id": "0", "to_type": "Video", "directed": false,
          "attributes": { "rating": 6.6, "date_time": 16000 }
        },
        {
          "e_type": "Video_AttributeTag", "from_id": "2", "from_type": "Video",
          "to_id": "action", "to_type": "AttributeTag", "directed": false,
          "attributes": { "weight": 1, "date_time": 0 }
        },
        {
          "e_type": "User_Video", "from_id": "2", "from_type": "VidUser",
          "to_id": "0", "to_type": "Video", "directed": false,
          "attributes": { "rating": 7.4, "date_time": 17000 }
        },
        {
          "e_type": "User_Video", "from_id": "0", "from_type": "Video",
          "to_id": "0", "to_type": "VidUser", "directed": false,
          "attributes": { "rating": 6.8, "date_time": 15000 }
        }
      ]
    }
  ]
}

Other versions of pathfinding algorithms are available in the GSQL Graph Algorithm Library.