Updated construct_api logic to parse host value #198
Conversation
lib/fluent/plugin/out_splunk_hec.rb
Outdated
| host = @hec_host.delete_prefix("/").delete_suffix("/").split("/", 2) | ||
| host_suffix = host[1].nil? ? "" : "/#{host[1]}" | ||
| URI("https://#{host[0]}:#{@hec_port}#{host_suffix}/services/collector") |
There was a problem hiding this comment.
it can only support one path, right? can it be updated so that it can support any number of it?
like https://mydomain.com/splunk/hec
There was a problem hiding this comment.
maybe a better approach would be adding another parameter. full_url.
use full_url if provided. if not build one using host, port, and protocol .
@chaitanyaphalak any thoughts?
There was a problem hiding this comment.
it can only support one path, right? can it be updated so that it can support any number of it? like
https://mydomain.com/splunk/hec
It will support any number of path as I am splitting host in 2 half only.
For mydomain.com/splunk/hec, it will split into mydomain.com and splunk/hec and will be join them with port value
There was a problem hiding this comment.
I like rock's idea
There was a problem hiding this comment.
@harshit-splunk i see. so it works.
hmm still, this way of parsing is not intuitive for the user.
instead of full_url we can add url_path parameter.
default would be "" but user can set it to /splunk/hec or anything.
However, having a full_url parameter is something that will benefit usability in the future. we can use it in the helm chart to reduce number of arguments that they need to pass.
this way, we are consistent with olly chart. (ie splunkPlatform.endpoint)
There was a problem hiding this comment.
@harshit-splunk
So I would like to suggest to edit this PR with full_url approach.
Resolving splunk/splunk-connect-for-kubernetes#665