CoWatchBenchmark-go

CoWatchBenchmark-go is a tool written by Go for benchmarking the performance of the cowatch server.

中文

Arguments

CoWatchBenchmark-go supports the following commands:

  -c string
        [Mandatory] configure: configure file
  -cr string
        [Mandatory] remote configure: remote configure. No coexistence with -c
  -v
        [optional] show verbos output

Configure file mode

The configure file is json style. The full content of the configure file as follows:

{
  "host": "http://localhost:8080",
  "room": 2,
  "user": 3,
  "msg_len": 1024,
  "msg_frequency": 10,
  "msg_random_send": 0,
  "log_enable": 0,
  "http_timeout": 25,
  "websocket_timeout": 45,
  "start_time_room": "2021-05-13T13:49:00.00+08:00",
  "start_time_user": "2021-05-13T00:10:00.00+08:00",
  "single_client_mode": 1,
  "parallel_mode": 2,
  "app_id": "abcdefghijklmn",
  "ws_request_speed_number": 100,
  "room_expiration_time": 300,
  "sdk_version": "1.0.0-8589-integration-a3d4cd01",
  "createRoomExtraData": {
    "expireTime": 5,
    "hostname": "co-test-golang",
    "maxMember": 4,
    "roomRole": "room_valid_until_expired",
    "hostLeaveRole": "host_transfer_to_second_after_host_leave",
    "ownerBackRole": "host_return_after_owner_back"
  },
  "http_header": {
    "Origin": "google.com",
    "user-agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/93.0.4577.82 Safari/537.36 Edg/93.0.961.52"
  }
}

parameter instruction

entry	type	default value	mandatory *	instruction
host	string		Y	address of coWatch server. schema://address
rooms	uint	1	N	the maximum number of room
users_per_room	uint	1	N	users in each room
message_frequency	uint	10/m	N	frequency of sending message per minute
message_length	uint	48 (bytes)	N	size of a message
http_request_timeout	uint	25(s)	N	http request timeout for create room
websocket_request_timeout	uint	25(s)	N	request timeout for websocket
parallel_mode	uint	0	N	mode for socket requesting server. 0 means serial 1 means parallel 2 mean batch
start_time_for_create_rooms	string		O	used in parallel = 1 only, such as "2021-05-13T13:49:00.00+08:00"
start_time_for_create_users	string		O	used in parallel = 1 only, such as "2021-05-13T13:50:00.00+08:00"
single_client_mode	uint	1	N	0: multi-point testing 1: single-point testing
app_id	string		Y	rtc token
ws_request_speed_number_for_mode_2	int		O	the number of ws request in a batch
room_expiration_time_in_second	uint	300s	Y	the living time for a room
sdk_version	string		Y	the sdk version
createRoomExtraData	json		Y	the custom data of body of posting when create the room. Note: the content in it can be empty
http_header	json		Y	the custom http header. Note: the content in it can be empty

• O in mandatory means options. It depends on other entries.

Note: Configure file mode will override other arguments

Local configure file

CoWatchBenchmark-go can be started with a configure file which locates in local driver. For example:

./CoWatchBenchmark -c ./config.json # config.json and CoWatchBenchmark are in the same directory.
# or
./CoWatchBenchmark -c /etc/cowatch/config.json

Remote configure file

CoWatchBenchmark-go can also be started with a remote configure file. It's very powerful when you want to getRooms multiple jobs in different client mechine. For example:

./CoWatchBenchmark -cr https://server_host/path/to/your/config

Using remote configure file can be start this program parallel in different client by setting the field start_time_room and start_time_user when you want to start multi-point jobs.

Testing mode

testing scale

From the perspective of test scale, there are two ways of single-point testing and multi-point testing.

• Single point : run one copy of this program
• Multi point : run multi copy of this program in multi host

request sequence

From the perspective of request sequence, it is divided into sequential requests, concurrent requests and batch requests.

• sequential requests: all http/ws requsets are sequential, i.e. one by one.
• concurrent requests: all http/ws requsets launch in the same time.
• batch requests: it alleviate the inefficiency of sequential requests. Essentially, it changes one request only at a time of sequential requests to several concurrent requests at a time.

Note: If the number of each batch in the batch requests is too large, the batch request will become a concurrent requests.

samples

Single point

"single_client_mode": 1

Multi point

"single_client_mode": 0

sequential requests

"parallel_mode":0

concurrent requests

"start_time_for_create_rooms": "2021-05-13T13:49:00.00+08:00",
"start_time_for_create_users": "2021-05-13T13:50:00.00+08:00",
"parallel_mode" : 1

batch requests

"parallel_mode" : 2,
"ws_request_speed_number_for_mode_2": 100

Extra parameter of creating room

There is a createRoomExtraData field in the config file. It supports adding extra data to the body of HTTP requesting. You should not add an array or an object into the createRoomExtraData object. That is to say, only string and number are supporting.

command line arguments mode

Refer to Arguments block