@Jdlrobson mentioned in T202786 that he successfully sent an mobile event while testing, so perhaps there are some validation issues going on as well. I saw some errors in the logs, but not nearly enough to explain this. @Ottomata, do you have any ideas by any chance?

I noticed that Schema.Edit is not being loaded on mobile.
The following code ( $resourceLoader->isModuleRegistered( 'schema.Edit' ) conditional ) is suspicious and I don't think it is running:

if ( ExtensionRegistry::getInstance()->isLoaded( 'EventLogging' ) ) {
        // schema.Edit is provided by WikimediaEvents
        if ( $resourceLoader->isModuleRegistered( 'schema.Edit' ) ) {
                  $schemaEdit += [
                        'dependencies' => [
                                'schema.Edit',
                                'mobile.startup'
                        ],
                        'scripts' => [
                                'resources/mobile.loggingSchemas/schemaEdit.js',
                        ]
                ];

The ResourceLoader module 'schema.Edit' is not being loaded by the client for some reason.
There were some changes to EventLogging by @Legoktm and @Krinkle around the time this went down, but I don't see anything obvious.

I've added code to MediaWiki:Mobile.js to make sure the schema is loading to see if the absence of this module is the problem:
https://en.m.wikipedia.org/wiki/Special:MobileDiff/859576064

That code looks problematic since schema modules are loaded in the same hook, and there's no guarantee that MobileFrontend's hook will run before EventLogging's does. Is there any reason MF can't use mw.track to lazy-load the schema like most other extensions now do?

In T203615#4585268, @Legoktm wrote:

That code looks problematic since schema modules are loaded in the same hook, and there's no guarantee that MobileFrontend's hook will run before EventLogging's does. Is there any reason MF can't use mw.track to lazy-load the schema like most other extensions now do?

I agree. This is likely legacy code that nobody bothered to update > T204584

Change 460949 had a related patch set uploaded (by Jdlrobson; owner: Jdlrobson):
[mediawiki/extensions/MobileFrontend@master] Remove conditional Edit event logging module.

https://gerrit.wikimedia.org/r/460949

In T203615#4585173, @Jdlrobson wrote:

I've added code to MediaWiki:Mobile.js to make sure the schema is loading to see if the absence of this module is the problem:
https://en.m.wikipedia.org/wiki/Special:MobileDiff/859576064

Unfortunately, this didn't work out—we didn't get any more events since that change.

select
    left(timestamp, 10) as hour,
    count(*)
from log.Edit_17541122
where
    event_platform = "phone" and
    timestamp >= "20180914"
group by left(timestamp, 10)

hour        count(*)
2018091404         1
2018091411         2
2018091422        29
2018091507         2

@Jdlrobson and I just decided that we'd sync up tomorrow to see whether events start coming in after the train deploy of his patch.

Change 460949 merged by jenkins-bot:
[mediawiki/extensions/MobileFrontend@master] Remove conditional Edit event logging module.

https://gerrit.wikimedia.org/r/460949

In T203615#4590990, @Neil_P._Quinn_WMF wrote:

In T203615#4585173, @Jdlrobson wrote:

I've added code to MediaWiki:Mobile.js to make sure the schema is loading to see if the absence of this module is the problem:
https://en.m.wikipedia.org/wiki/Special:MobileDiff/859576064

Unfortunately, this didn't work out—we didn't get any more events since that change.

Reminder to revert the above if/when it isn't needed. Haven't confirmed, but likely causes extra overhead early during the page load process. It might also hinder the ability to batch the requests for schema modules due to one being demanded separately before the batch is discovered.

In T203615#4591459, @Krinkle wrote:

Reminder to revert the above if/when it isn't needed. Haven't confirmed, but likely causes extra overhead early during the page load process. It might also hinder the ability to batch the requests for schema modules due to one being demanded separately before the batch is discovered.

Jon did this yesterday, for the record :)