Change Details

It would be nice if Kartotherian when calling WDQS on behalf of the client would identify which client requested it. Since WDQS is an internal WMF service, there should be no PII proliferation concern about it, and we could have more efficient load controls and throttling for it. Possible ways, in order of preference: * X-Client-IP (unless varnish would kill it) * Adding client IP to the User Agent * Adding another header * Adding client IP to request string as extra parameter * something else?

It would be nice if Kartotherian when calling [[ https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service | Wikidata Query Service (WDQS) ]] on behalf of the client would identify which client requested it. Since WDQS is an internal WMF service, there should be no PII proliferation concern about it, and we could have more efficient load controls and throttling for it. Possible ways, in order of preference: * X-Client-IP (unless varnish would kill it) * Adding client IP to the User Agent * Adding another header * Adding client IP to request string as extra parameter * something else?

It would be nice if Kartotherian when calling [[ https://www.wikidata.org/wiki/Wikidata:SPARQL_query_service | Wikidata Query Service (WDQS) ]] on behalf of the client would identify which client requested it. Since WDQS is an internal WMF service, there should be no PII proliferation concern about it, and we could have more efficient load controls and throttling for it. Possible ways, in order of preference: * X-Client-IP (unless varnish would kill it) * Adding client IP to the User Agent * Adding another header * Adding client IP to request string as extra parameter * something else?