https://bugzilla.wikimedia.org/show_bug.cgi?id=53806

--- Comment #5 from [email protected] ---
Looking further, we do not log the X-Analytics header of the request, but
the X-Analytics header of the response. Could it be that the caching
interferes here?

Doing some quick check for the 404-01 carrier (India) seems to
somewhat support this, as ~30% of hits come misattributed, while only
~1% of the misses do. It a least shows a really heavy bias that hits
are more affected than misses.

Could it be (wild guess) that upon a cache hit, varnish uses the
X-Analytics header of the cached document?





If the caching is not the issue... I am not sure if this is related,
but X-Analytics column in the logs is not always
  zero=XXX-XX
We also have log lines where the X-Analytics column
  zero=XXX-XX;mf-m=a
  zero=XXX-XX;mf-m=b
and in the sampled-1000 (not limited to mobile or zero) stream we also
see some
  mf-m=a
  mf-m=b
. I could not find this 'mf-m' in the puppet repo. And each of those
'mf-m' log lines is for a 'pass/200' request.

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are on the CC list for the bug.
_______________________________________________
Wikibugs-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikibugs-l

Reply via email to