When I was working on related stuff, I found that the value of x_analytics_map ia null on the wmf.webrequest table in stat1002, when is_pageview is filtered for true, and agent_type is user. I'm wondering why that would be.
These are the things I found:
For 28th April 2015, of 741,858,511 requests, 28,827,374 have x_analytics(is set to '-') and x_analytics_map set to null. It's about 3.9% of all requests that day.
You can find these counts by doing something like this on hive in the server. - SELECT count(*) FROM webrequest WHERE x_analytics_map IS NULL AND agent_type = 'user' AND is_pageview = TRUE AND YEAR = 2015 AND MONTH = 4 AND DAY = 28;
Does anyone have ideas on why this might be and if something underlying is broken?
--Madhu :)
Well, if a request is http-only, on desktop, and not through zero, what x_analytics value does it have? None.
The mapper consumes x_analytics, extracts key-value pairs, and turns that into a map. So if there are no key-value pairs, the map is NULL.
On 29 April 2015 at 17:06, Madhumitha Viswanathan mviswanathan@wikimedia.org wrote:
When I was working on related stuff, I found that the value of x_analytics_map ia null on the wmf.webrequest table in stat1002, when is_pageview is filtered for true, and agent_type is user. I'm wondering why that would be.
These are the things I found:
For 28th April 2015, of 741,858,511 requests, 28,827,374 have x_analytics(is set to '-') and x_analytics_map set to null. It's about 3.9% of all requests that day.
You can find these counts by doing something like this on hive in the server. - SELECT count(*) FROM webrequest WHERE x_analytics_map IS NULL AND agent_type = 'user' AND is_pageview = TRUE AND YEAR = 2015 AND MONTH = 4 AND DAY = 28;
Does anyone have ideas on why this might be and if something underlying is broken?
--Madhu :)
Analytics mailing list Analytics@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/analytics