Newer
Older
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
# Task Service - Task Documentation
### Task Definition
A tasks is described in JSON format and base task (template) definition *must* be available
before a task can be created. Task JSON templates are stored in Mongo database in a collection
named `taskTemplates`. Some task properties are statically predefined in the template,
others are populated dynamically by the input request and the output response. Below is
an example of task template definition:
```json
{
"name":"exampleTask",
"url":"https://jsonplaceholder.typicode.com/todos/1",
"method":"GET",
"requestPolicy":"example/example/1.0",
"responsePolicy":"",
"finalPolicy":"",
"cacheNamespace":"login",
"cacheScope":"user"
}
```
Tasks are created by using their `name` attribute. If a task template with the given
`name` does not exist, a task will not be created and an error is returned.
### Task Execution
Below is an example of creating a task with the template definition given above:
```shell
curl -v -X POST http://localhost:8082/v1/task/exampleTask -d '{"exampleInput":{"test":123}}'
```
The HTTP request will create a task for asynchronous execution and the JSON object
given as input will be used as the body of the task request. The caller will receive
immediately the `taskID` as response, and the result of the asynchronous task
execution will be stored in the TSA Cache after the task is completed.
### Task Executor Configuration
There are two environment variables that control the level of concurrency
of the task executor.
```shell
EXECUTOR_WORKERS="5"
EXECUTOR_POLL_INTERVAL="1s"
```
Poll interval specifies how often the executor will try to fetch a *pending* task
from the queue. After a task is fetched, it is given to one of the workers for execution.
With the given example of 1 second (1s), the executor will retrieve 1 task per second at most.
If there are multiple instances (pods) of the service, multiply by their number
(e.g. 5 pods = 5 tasks per second).
If this is not enough, the poll interval can be decreased, or we can slightly modify
the polling function to fetch many tasks at once (and also increase the number of workers).
To learn more about the queue and why we use database as queue see [queue](queue.md)
### Task Storage
We use MongoDB for tasks storage. There are three Mongo collections with different purpose.
1. **taskTemplates**
The collection contains predefined task definitions in JSON format. Here are defined
what tasks can be created and executed by the service.
2. **tasks**
The collection contains newly created tasks *pending* for execution. It acts like a
FIFO queue and is used by the task executor to retrieve tasks for workers to execute.
3. **tasksHistory**
The collection contains successfully completed tasks for results querying,
audit, reporting and debugging purposes.