Closed Bug 1056542 Opened 10 years ago Closed 10 years ago

Export telemetry data for CHARSET_OVERRIDE_SITUATION by locale for Firefox 30

Categories

(Mozilla Metrics :: Data/Backend Reports, defect)

defect
Not set
normal

Tracking

(Not tracked)

RESOLVED FIXED
Unreviewed

People

(Reporter: hsivonen, Unassigned)

Details

Attachments

(1 file)

Could you, please, run the analysis script used in bug 965707 with the following two modifications: 1) Looking for CHARSET_OVERRIDE_SITUATION instead of CHARSET_OVERRIDE_USED 2) Instead of 2 columns for values 0 and 1, have 8 columns for values 0 through 7 for Firefox&Fennec 30 on the release channel from 2014-06-30 to 2014-07-20, inclusive?
Group: metrics-private
Do you want to try using the Telemetry Self-Serve analysis framework to run this yourself? I can help with the code changes and give you the code I used to run the analysis in bug 965707.
(In reply to Mark Reid [:mreid] from comment #1) > Do you want to try using the Telemetry Self-Serve analysis framework to run > this yourself? I can help with the code changes and give you the code I > used to run the analysis in bug 965707. If you can point me to documentation on how to use the Self-Serve framework, I can try. Currently I don't know how. Thanks.
I was looking at making the modifications to the code to use _SITUATION instead of _USED, and checked the Telemetry dashboard. It looks like there are no submissions for the CHARSET_OVERRIDE_SITUATION histogram for the given period (2014-06-30 to 2014-07-20) http://mzl.la/1lHKKGh
(In reply to Mark Reid [:mreid] from comment #3) > I was looking at making the modifications to the code to use _SITUATION > instead of _USED, and checked the Telemetry dashboard. It looks like there > are no submissions for the CHARSET_OVERRIDE_SITUATION histogram for the > given period (2014-06-30 to 2014-07-20) > > http://mzl.la/1lHKKGh Oops, don't mind me - that's being shown by "build date" and not by calendar date. Changing the toggle to show by calendar date shows the data is there as expected.
Would you like to see the individual submissions, or should I combine the counts for a given locale? For example, given this data: Locale App Ver Histogram Count ------ ------- ------- ---------------------------- ----- de Firefox 30.0 [0, 0, 0, 0, 1, 1, 0, 0, 0] 8 de Firefox 30.0 [0, 0, 0, 0, 1, 2, 0, 0, 0] 2 de Firefox 30.0 [0, 0, 0, 0, 3, 0, 0, 0, 0] 1 Would you like it output as above, or would you prefer to see a single line of: de Firefox 30.0 [0, 0, 0, 0, 13, 12, 0, 0, 0] Where 13 = 8 x 1 + 2 x 1 + 1 x 3 12 = 8 x 1 + 2 x 2
Flags: needinfo?(hsivonen)
(In reply to Mark Reid [:mreid] from comment #5) > Would you like it output as above, or would you prefer to see a single line > of: The uncombined form is actually useful for figuring out which reasons are the ones that the user re-overrides, so now that I think about it, I'd prefer the uncombined version. (I'll need combined, too, but I can compute that myself in post-processing.) Thanks.
Flags: needinfo?(hsivonen)
Attached file bug1056542-0.1.tar.gz
** Running the CHARSET_OVERRIDE_SITUATION analysis Further details at mreid's blog @ http://mreid-moz.github.io/blog/2013/11/06/current-state-of-telemetry-analysis - Sign in to telemetry-dash.mozilla.org with an @mozilla.com email address - Click "Launch an ad-hoc analysis worker" - Give it a name (such as "hsivonen-bug1056542") - Add your openssh key (~/.ssh/id_rsa.pub) - Click "Submit". The page will update until the server is ready, so just wait until the server's public hostname appears (plus a couple of minutes for the server to configure itself). You should also receive a confirmation email that the server has launched. I've created some helper scripts to run the analysis, they're attached as a tarball. Copy this over to the server via scp. Also copy the "run_many.sh" script over. local$ export HOST=ubuntu@ec2-BLA.us-west-2.compute.amazonaws.com local$ tar xvf bug1056542-0.1.tar.gz run_many.sh local$ scp bug1056542-0.1.tar.gz $HOST:~/ local$ scp run_many.sh $HOST:/mnt/telemetry/ local$ ssh $HOST worker$ cd /mnt/telemetry worker$ bash run_many.sh 30 20140630 20 This will generate a bunch of files in the /mnt/telemetry/output directory, one for each day of data. It will take some time - if you want to speed it up, you can launch several telemetry worker instances and run a subset of the days on each one in parallel. worker1$ bash run_many.sh 30 20140630 6 worker2$ bash run_many.sh 30 20140707 6 worker3$ bash run_many.sh 30 20140714 6 Download the output files once things are complete, but make sure not to wait more than 24 hours, since these workers are killed automatically after 24 hours.
(In reply to Mark Reid [:mreid] from comment #7) > Download the output files once things are complete, but make sure not to wait > more than 24 hours, since these workers are killed automatically after 24 > hours. Thank you! I'll proceed early next week in order to avoid failing at the 24-hour limit due to the weekend.
Thanks. I got the data out of the system by following the instructions. Now I need to analyze it.
Status: NEW → RESOLVED
Closed: 10 years ago
Resolution: --- → FIXED
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: