How to print the client's original request URL in the logs instead of the rewritten URL

[wuzhuoquan]

Title: One line description

Description:
I wanted to print the client's original request URL in the logs, so I used request_url: '%REQ(X-ENVOY-ORIGINAL-PATH?:PATH)%' in logFormat to output it to the logs. Later, I discovered that the X-ENVOY-ORIGINAL-PATH header was also being forwarded to the server. In cases where the URL contains many parameters or is particularly long, this could cause the header to become too large and result in a 431 error. Therefore, I used suppress_envoy_headers to remove this header.
However, after making this change, I noticed that the request_url field in the logs no longer shows the original path from the client request, but instead shows the path after Envoy's rewrite. For example, when a client requests /payment/v1/list, it gets rewritten to /payment/list before being sent to the server, and at this point the printed request_url is /payment/list instead of /payment/v1/list. This behavior is very different from Nginx. Could you please help me how to print the client's original path while still using suppress_envoy_headers to remove the X-ENVOY-ORIGINAL-PATH header? Thank you.

[RyanTheOptimist]

cc: @auni53 @zuercher

[agrawroh]

One way to do this is by using H2M to write the value of this header to metadata, remove the header so that it does not get propagated to the backend server and then use the metadata to populate your access logs.

For example, you can do:

http_filters:
  - name: envoy.filters.http.header_to_metadata
    typed_config:
      "@type": type.googleapis.com/envoy.extensions.filters.http.header_to_metadata.v3.Config
      request_rules:
        - header: x-envoy-original-path
          on_header_present:
            metadata_namespace: metadata.incoming_request
            key: original_path
            type: STRING
          remove: true
        - header: :path
          on_header_present:
            metadata_namespace: metadata.incoming_request
            key: path
            type: STRING
          remove: false
...

And then in access logs, you can use %DYNAMIC_METADATA(metadata.incoming_request:original_path)% instead of the request header.

[github-actions]

This issue has been automatically marked as stale because it has not had activity in the last 30 days. It will be closed in the next 7 days unless it is tagged "help wanted" or "no stalebot" or other activity occurs. Thank you for your contributions.