It looks like we're working with you on your particular case via email, but I'll address the general issue here.
First, what we're doing about it. Development has been working on building a brand-new uploader tool from scratch using modern client-side upload methods, which will also send files directly to our storage provisions on Amazon S3 (rather than caching files locally first before syncing them to S3, as we do currently), so among numerous other benefits this will eliminate any contributing factors related to Apache's implementation of the ancient HTTP upload spec and our other server-side tools which interface with it.
Meanwhile for the time being, our turnaround on the alternate upload method is normally same-day, but lately Development had been performing these personally in order to analyze uploads in progress in hopes of finding some contributing factor or mitigation we could apply, which extended the turnaround time. As a result of this, they did recently relax a network setting to allow a longer timeout period before idle connection sessions get closed, and this seems to have eliminated the issue in many cases.
The recurring problem stalling our current upload tool has affected fewer than one percent of our users for at least a year now, albeit only sporadically. Whenever the problem recurs, it later spontaneously resolves as mysteriously as it began, only to crop up again a few weeks or months later, despite no changes at our end that could account for it -- and we've certainly been applying every method at our disposal to identify any possible contributing factor or mitigation strategy within our control.
Most affected users resolve the problem by using a different browser, OS, computer, or ISP connection, but a tiny handful of users such as yourself seem to remain affected persistently despite changing every possible variable at their end, which suggests a critical contributing factor in such cases may reside with a key backbone provider serving ISPs in their particular geographic area.
