Tugraph Running

This document describes the operation mode of the TuGraph service, starting, stopping, and restarting operations.And TuGraph’s service configuration parameters, configuration file format, and command-line configuration parameters.

1. Prerequisites

TuGraph has the following prerequisites for its execution: a correct installation of TuGraph.

TuGraph execution requires ensuring that the library file “liblgraph.so” is located in the environment variable LD_LIBRARY_PATH.

The user running the TuGraph process does not need superuser privileges. However, they need read permissions for the configuration file (usually lgraph.json) and any files referenced within it. Additionally, they should have write permissions for data folders, log files, etc.

2.The operation mode

TuGraph can be started as a normal process in the foreground or as a daemon in the background.

When running as a normal process, TuGraph can print logs directly to the terminal, which is handy when debugging server configurations. However, because the foreground process is terminated after the terminal exits, the user must ensure that the terminal remains open while the TuGraph server is running. In daemon mode, on the other hand, the TuGraph server can continue to run even if the terminal that started it exits. If TuGraph needs to run for a long time, it is recommended to start the TuGraph server in daemon mode.

2.1.Running a normal process

The ‘lgraph_server -d run’ command runs TuGraph as a normal process. Normal processes depend on the command line terminal, so when the terminal ends, the TuGraph process is automatically terminated. Normal process mode with ‘–log_dir “”’ can output the process log directly to the terminal, so it is easier to debug.

Start the command：

$ ./lgraph_server -c lgraph_standalone.json --log_dir ""

Example of running output in normal mode：

20200508120723.039: **********************************************************************
20200508120723.039: *                  TuGraph Graph Database v3.1.0                     *
20200508120723.040: *                                                                    *
20200508120723.041: *        Copyright(C) 2018 Ant Group. All rights reserved.           *
20200508120723.041: *                                                                    *
20200508120723.044: *             Licensed host: hostname      threads:0, ha:0           *
20200508120723.044: **********************************************************************
20200508120723.044: Server is configured with the following parameters:
20200508120723.045:   data directory:    ./lgraph_db
20200508120723.045:   license:           ./fma.lic
20200508120723.046:   enable ha:          0
20200508120723.046:   async:              0
20200508120723.047:   host:               127.0.0.1
20200508120723.047:   REST port:          7071
20200508120723.048:   RPC port:           9091
20200508120723.048:   enable rpc:         0
20200508120723.051:   optimistic txn:     0
20200508120723.059:   verbose:            1
20200508120723.074:   log_dir:
20200508120723.074:   ssl_auth:           0
20200508120723.075:   resource dir:       ./resource

20200508120723.077: Loading DB state from disk
20200508120723.110: [RestServer] Listening for REST on port 7090
20200508120723.110: [LGraphService] Server started.

In normal process mode, the user can prematurely terminate the TuGraph process by pressing ‘CTRL+C’.

2.2.Run process daemon mode

Start the command：

$ ./lgraph_server -d start -c lgraph_daemon.json

Example output from running in daemon mode:

Starting lgraph...
The service process is started at pid 12109.

The TuGraph server process started by this command is a daemon process that loads the relevant configuration from the file ‘lgraph_daemon.json’. After the server starts, it will start printing logs in a log file that can then be used to determine the status of the server.

3.Service operation

3.1.Start the service

TuGraph needs to be started using the ‘lgraph_server -d start’ command line. The following is an example of the command to start TuGraph:

$ ./lgraph_server -d start -c lgraph_daemon.json
Starting lgraph...
The service process is started at pid 12109.

The TuGraph server process started by this command is a daemon process that loads the relevant configuration from the file ‘lgraph_daemon.json’. After the server starts, it will start printing logs in a log file that can then be used to determine the status of the server.

3.2.Stop the service

You can stop the TuGraph daemon using the ‘kill’ command and the ‘lgraph_server -d stop’ command. Since it is possible to run multiple TuGraph server processes on the same computer, we distinguish between the different server processes using the ‘.pid ‘file, which is written to the working directory where the process was started. Therefore, you need to run the command ‘lgraph_server-d stop’ in the same working directory to stop the correct server process.

user@host:~/tugraph$ ./lgraph_server -d start -c lgraph_standalone.json
20200508122306.378: Starting lgraph...
20200508122306.379: The service process is started at pid 93.

user@host:~/tugraph$ cat ./lgraph.pid
93

user@host:~/tugraph$ ./lgraph_server -d stop -c lgraph_standalone.json
20200508122334.857: Stopping lgraph...
20200508122334.857: Process stopped.

3.3.Restart the service

Users can also restart the TuGraph service by ‘lgraph_server -d restart’ :

$ ./lgraph_server -d restart
Stopping lgraph...
Process stopped.
Starting lgraph...
The service process is started at pid 20899.

4.Service configuration

The TuGraph server loads configurations from the configuration file and command line options at startup, and if different values are specified for the same option in the configuration file and command line, the value specified in the command line will be used preferentially.

4.1.Configuration parameters

The parameters and their types are described as follows:

Parameter names	Type	Instructions
directory	string	Directory where data files are stored. If the directory does not exist, it is automatically created. The default directory is /var/lib/lgraph/data.
durable	boolean	Whether to enable real-time persistence. Turning off persistence can reduce the disk IO overhead when writing, but data may be lost in extreme cases such as machine power failure. The default value is `true`.
host	string	The IP address on which the REST server listens. The default address is 0.0.0.0.
port	int	The Port on which the REST server listens. The default port is 7070.
enable_rpc	boolean	Whether to use RPC services. The default value is 0.
rpc_port	int	Port used by RPC and HA services. The default port number is 9090.
enable_ha	boolean	Whether to enable the HA mode. The default value is false.
ha_log_dir	string	HA log directory. The HA mode needs to be enabled. The default value is null.
verbose	int	Detail level of log output information. The value can be 0,1,2. The larger the value, the more detailed the output information. The default value is 1.
log_dir	string	Directory where log files are stored. The default directory is /var/log/lgraph/.
ssl_auth	boolean	Whether to use SSL authentication. When HTTPS is enabled, only the HTTPS service is enabled on the REST server. The default value is false.
web	string	The directory where the web files (including the visual part) are located. The default directory is/usr/local/share/lgraph/resource.
server_cert	string	certificate file path used by the server when SSL authentication is enabled. The default path is/usr/local/etc/lgraph/server - cert.pem.
server_key	string	The public key file used by the server when SSL authentication is enabled. The default directory is/usr/local/etc/lgraph/server - key.pem。
enable_audit_log	boolean	Whether to enable audit logs. The default value is false.
audit_log_expire	int	When audit logs are enabled, the validity period (hours) of audit logs is automatically cleared upon timeout. If the value is 0, the logs are not cleared. The default value is 0.
audit_log_dir	string	Specifies the directory for storing log files when audit logs are enabled. The default directory is $directory/audit_log.
load_plugins	boolean	Import all stored procedures when starting the service. The default value is 1.
optimistic_txn	boolean	Enable optimistic multithreaded write transactions for Cypher. The default is 0.
disable_auth	boolean	Disable REST authentication. The default is 0.
ha_snapshot_interval_s	int	Snapshot interval (in seconds) The default value is 86400.
ha_heartbeat_interval_ms	int	Heartbeat interval in milliseconds. The default value is 1000.
ha_node_offline_ms	int	The heartbeat times out and the interval (in milliseconds) between nodes going offline. The default value is 60000.
ha_node_remove_ms	int	The interval (in milliseconds) at which a node is considered completely dead and removed from the list. The default value is 120000.
enable_ip_check	boolean	Allow IP address whitelists. The default value is 0。
idle_seconds	int	The maximum number of seconds a child process can be idle. The default value is 600.
enable_backup_log	boolean	Whether to enable backup logging. The default value is 0.
backup_log_dir	string	The directory where backup files are stored. The default value is null.
snapshot_dir	string	Directory where snapshot files are stored. The default value is null.
thread_limit	int	The maximum number of threads that can be used simultaneously. The default value is 0, which means that no restriction is imposed. The license file prevails.
unlimited_token	boolean	Whether to set the link token to be infinite. The default value is 0 and the validity period is 24 hours.
enable_fulltext_index	boolean	Whether to enable the full-text index function, the default value is 0.
fulltext_analyzer	string	Full-text index tokenizer type. Can be set to `StandardAnalyzer` or `SmartChineseAnalyzer`. Default is `StandardAnalyzer`
fulltext_commit_interval	int	Full-text index data submission period, for write operations, in seconds. The default is 0, which commits immediately.
fulltext_refresh_interval	int	Full-text index data refresh cycle, for read operations, in seconds. The default is 0, and the latest written data can be read immediately.
conf	string	Initialize the HA cluster according to host1:port1,host2:port2,host3:port3, the default value is empty.
ha_node_join_group_s	int	The waiting time for a node to try to join the high availability cluster, in seconds, the default is 10.
ha_bootstrap_role	int	Whether to use bootstrap to start, and the server role to start using this method, 0 means not to use bootstrap to start, 1 means to use bootstrap to start and this node is a leader, 2 means to use bootstrap to start and this node is a follower, only these 3 options. The default value is 0.
help	boolean	Print the help message. The default value is 0.

4.2.Server configuration file

TuGraph’s configuration file is stored in JSON format. It is recommended that most configuration be stored in configuration files and that some configuration parameters be modified only temporarily using command-line options when needed.

A typical configuration file looks like this:

{
  "directory": "/var/lib/lgraph/data",
  "license": "/var/lib/lgraph/fma.lic",

  "port": 7090,
  "rpc_port": 9090,
  "enable_ha": false,

  "verbose": 1,
  "log_dir": "/var/log/lgraph/",

  "ssl_auth": false,
  "server_key": "/usr/local/etc/lgraph/server-key.pem",
  "server_cert": "/usr/local/etc/lgraph/server-cert.pem"
}