I would probably recommend using the existing
EventLogging infrastructure
for sending the data to our back end, assuming it won't explode under heavy
load spikes... Which it might. :)
Eventlogging is not the best choice. Besides
not handling bursts of traffic it is -currently- a tier-2
service. Any logging infrastructure should be
tier-1.
http://m.mediawiki.org/wiki/EventLogging/OperationalSupport
On Jul 25, 2014, at 1:11 AM, Brion Vibber <bvibber(a)wikimedia.org> wrote:
I would probably recommend using the existing
EventLogging infrastructure
for sending the data to our back end, assuming it won't explode under heavy
load spikes... Which it might. :)
>
> -- Brion
> On Jul 24, 2014 3:14 PM, "Gergo Tisza" <gtisza(a)wikimedia.org> wrote:
>
>> Hi all,
>>
>> frontend development is greatly hindered by not having logs of errors that
>> happen in production. If there is a mistake in a PHP file, it is usually
>> quickly caught after deployment when a large number of exceptions show up
>> in the error log. If the mistake is in a JS file, it can take a long time
>> until the error is reported and reproduced; especially so if it only
>> happens under exotic conditions.
>>
>> Many sites solve this issue by setting up an error handler in Javascript
>> which reports any errors that occurred to a logging server. I tried to make
>> a laundry list of things that need to be done or considered if we want to
>> set up such logging for Wikimedia sites and/or MediaWiki in general; I put
>> it up as a draft RfC at [1]. I would appreciate feedback on whether this is
>> plausible or worthwhile.
>>
>> [1]
>>
>>
https://www.mediawiki.org/wiki/Requests_for_comment/Server-side_Javascript_…
>> _______________________________________________
>> Wikitech-l mailing list
>> Wikitech-l(a)lists.wikimedia.org
>>
https://lists.wikimedia.org/mailman/listinfo/wikitech-l
> _______________________________________________
> Wikitech-l mailing list
> Wikitech-l(a)lists.wikimedia.org
>
https://lists.wikimedia.org/mailman/listinfo/wikitech-l