Web browser: ---
            Bug ID: 60314
           Summary: Requests through ssl terminators hardly present in
                    edits tsvs before 2014-01-14
           Product: Analytics
           Version: unspecified
          Hardware: All
                OS: All
            Status: NEW
          Severity: normal
          Priority: Unprioritized
         Component: General/Unknown
    Classification: Unclassified
   Mobile Platform: ---

It seems fixing bug 58876 made the logged requests from ssl
terminators jump from ~50/day to 1.3M/day [1] ... which is actually good.

As the ~1.3M/day seems more plausible than ~50/day ... Did some of the
(maybe) relevant changes
really unbreak edit.tsv writing for ssl terminators as a side-effect?

If so, how could they have unbroken storing requests from ssl
terminators ... should not x-encoded requests have gotten logged
(They for sure were in the stream before, as I checked before writing
the changes above, and some are visible in the edit tsvs for the
previous dates.)

Was there some network issue beforehand?
Was udp-filter (see [4] for its invocation) misbehaving?

Once we understand what went on, would we need fixes elsewhere?

Is the 1.3M/day now finally about correct?

Some more pointers:

* The above gerrit changes got deployed to ssl1001 on 2014-01-14
  ~18:56 [2].

* Heavy ssl traffic in the edit tsvs starts on 2014-01-14T18:59:11.634
  and is limited to ssl1001 at first. On 2014-01-14T19:06:23.209 other
  machines kick in.

* There was a deployment window on 2014-01-14 19:00-21:00 upgrading
  group1 1.23wmf9->1.23wmf10 [3], but it looks unrelated, as we're
  seeing ssl requests pour in for unaffected wikis.

qchris@stat1002 // 0 // 22:13:50                                  
cwd: ~
zgrep -c '^ssl' /a/squid/archive/edits/edits.tsv.log-201401*.gz 

[2] See
around 18:54, look for ottomata & qchris.



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

Reply via email to