Du ser en gammel version af denne side. Se den nuværende version.

Sammenlign med nuværende Vis sidehistorik

« Forrige Version 2 Næste »

If we want to measure and log "real" pageviews from out Confluence, see Access Logging in Confluence, and we a not doing this from within Confluence, we need to filter out false results.

this is my current filterring Results:

Robots and Crawlers

We need to filter out those:

Kibana / Grafana
type: apache AND (request.keyword: \/display* OR request.keyword: \/pages\/viewpage.action*) AND NOT ( agent: *facebook* OR agent: *crawler* OR agent: *bot* OR agent: *Bot* OR agent: *Spider*) 
Splunk
index=apache AND host=moserver AND (uri="*/display/*" OR uri="*/viewpage.action/*") useragent!="*bot*" useragent!="*spider*"useragent!="*facebookexternalhit*" useragent!="*crawler*" 

 

HTTP Codes

Often, if You page is down or under maintenanace, apache/ngix logs will return HTTP Code 502 or HTTP Code 503 - See https://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html

These can be removed from the search results; - or HTTP Code 200 can be added to give only possitive results:

Kibana / Grafana
type: apache AND (request.keyword: \/display* OR request.keyword: \/pages\/viewpage.action*) AND response: 200 AND NOT ( agent: *facebook* OR agent: *crawler* OR agent: *bot* OR agent: *Bot* OR agent: *Spider*) 
Splunk
index=apache AND host=moserver AND (uri="*/display/*" OR uri="*/viewpage.action/*") useragent!="*bot*" useragent!="*spider*"useragent!="*facebookexternalhit*" useragent!="*crawler*"  AND status=200

Reffers

A User request for a Confluence page spawns several request from the Confluence to the Confluence itself, and from the client to the Confluence server, but only one represents the actual page request. In the above, this is covered by:

Kibana / Grafana
request.keyword: \/display* OR request.keyword: \/pages\/viewpage.action*
Splunk
uri="*/display/*" OR uri="*/viewpage.action/*"

Monitoring

Monitoringring on a website from various Souces (Zenoss, Nagios, Datadog etc etc) also represents a large (often the largest) number of hits, and we need to remove these, this can be done is several ways:

  • Identifying all agent types of the monitoring
  • Often monitoring is on / - giving the agent a HTTP Code 302 back

Looking into You logs will often make it pretty easy to Identify monitoring:

 

 

 

 

  • Ingen etiketter