how to execute 2 branch components concurrently

[lizhe19]

input:
generate:
mapping: root = {"reqCode":"123456","data":"info"}
interval: 0s
count: 1
pipeline:
processors:
- mapping: |
meta start = timestamp_unix_milli()
- log:
level: INFO
message: 'once requet record start time：${!now().ts_format("2006-01-02 15:04:05")}'
- parallel:
cap: 0
processors:
- branch:
processors:
- http:
headers:
Content-Type: application/json; charset=UTF-8
TEST: test
url: http://172.32.6.25:9357/test/http2
verb: POST
timeout: 30s
parallel: true
- catch:
- log:
level: ERROR
message: "request execption,request body=${!content()}"
- branch:
processors:
- http:
headers:
Content-Type: application/json; charset=UTF-8
TEST: test
url: http://172.32.6.25:9357/test/http3
verb: POST
timeout: 30s
parallel: true
- catch:
- log:
level: ERROR
message: "request execption,request body=${!content()}"
- log:
level: INFO
message: 'once request used time total：${!(timestamp_unix_milli() - metadata("start"))}ms'
output:
stdout: {}

[lizhe19]

i can not found any answer from the help doc, i try the "parallel" component, but can not take effect. how should i do?

[mihaitodor]

Hey @lizhe19, messages need to be batched if you want the parallel processor to have any effect. Since you have your processors in the pipeline section, they'll be executed in parallel according to pipeline.threads (see docs here). However, I guess you have a single message in there, given that generate.count is set to 1 (your sample yaml is not well formatted). You can set that to a higher value and get rid of the parallel processor. Alternatively, if you do have some other input which uses batching (or if you set generate.count and generate.batch_size to a values bigger than 1), you can make use of the parallel processor as well, but note that you'd get 2 levels of parallelism in that case if your processors are within the pipeline section.

LE: Transferring this to a discussion as per #2026.

[lizhe19]

你好...谢谢你的邮件 我已收到！