Issue

Christopher
Technical Support
StableBit CloudDrive
not sure, latest 1.0.0.479
Public
Alex

Yes, funny that this came in, I was just thinking about optimizing that scenario.

That looks "normal" to me. There are times when a chunk will need to be uploaded multiple times due to a change on the disk, especially if that chunk holds on-disk metadata. So it's not really getting "stuck", there are just a whole bunch of updates queued up for that chunk. It's also possible that something is constantly updating that part of the disk and so CloudDrive keeps sending up those updates to the cloud.

CloudDrive does combine multiple chunk uploads into one, and you can actually see it doing that here (if you hover your mouse over those states for an explanation). But in this case it's only combining 2 requests at a time, because there are only 2 upload threads in this case. If you increase your upload threads, you will see a whole bunch of writes get combined at once (if they're in the queue).

Now, I was just  thinking that in this case we should probably boost the number of write threads temporarily in order to combine more write requests into one. In fact, maybe the I/O manager should always keep a minimum of 10 threads and then the provider I/O should do the actual limiting based on the I/O performance settings.

I'm going to try that and see if I can make that work.

Also, perhaps we should have a "wait until X amount of data needs to be uploaded before uploading anything" option, in order to prevent this scenario? That may work too.

So thank you for the feedback, and yes, I'm going to address this soon one way or another.