Uploading files use a much download bandwith

[Jean-Michel]

Hello

I'm trying to use Syncovery 10.15.10 in a docker with Ubuntu 24.04, on a Core i5, 4 CPU, 8GB, HDD (non SAT) to upload files from external, USB3 connected SSD drives to Amazon S3
For now I have 2 SSD connected on 2 different USB ports
2 jobs running, transferring 2 files per SSD each, so a total of 4 files simultaneously
Average upload speed if 50-55 MBps in total over a 1Gbps consumer fiber Internet access

So far so good, happy with the performance
But I am surprised that my router tells me that I am using as much download bandwith as upload.

I wonder if Syncovery is downloading what it has upload for verification ?
Which sounds surprising because as far as I remember, you can't access an object in S3 until the upload is complete.

See bandwidth diagrams below (red is upload, green is download)
Any hint if there's an option in Syncovery that I can disable ?
The problem is that if syncovery is re-downloading what we are uploading, it creates a huge e-gress cost with AWS

Thanks

[tobias]

Hello,
the only explanation could be if you have chosen "Verify copied files" on the "Files" tab sheet in the profile. For smaller files, Syncovery can use hashes, but for larger files, it has to re-download them to compare.

Maybe you can try pausing the jobs to see if the download bandwidth use goes down, or if the downloads are done by a different piece of software.

If problems remain, you could maybe post the profile settings or send a log file to support@syncovery.com.

[Jean-Michel]

Hi Tobais

Thanks for your fast response
I don't have "Files" -> "Files" -> "Verify copied files" checked but I have that other "Files" -> "More" -> "Check destination file (again)" checked
The seems to be default, I didn't changed any of those

What I find strange is that each file is large and takes at least 1 hour to be uploaded
And in AWS S3, until the file upload is complete you are not supposed to be able to read/download it.
So any verification/read-back could only occur shifted by ~1 hour in time.
But on the bandwidth graph, the download bandwidth starts exactly at the same time as the upload bandwidth. And when the job ends, they both ends at the same time.

May be it's an artifact of the AWS SDK used.

Unless you have other ideas, I will test with other tools like the AWS CLI, s5cmd to see if the same artifact occurs.

Thanks

[tobias]

Hello,
I still can't explain it. Maybe the downloads are completely unrelated to Syncovery? Maybe somebody in your house is downloading something? Or some other job runs somewhere.

We don't use any AWS SDK.

You could send a log file to support@syncovery.com.

If you could create a smaller test case, you could turn on Internet Protocol Logging on the Program Settings dialog, tab sheet Logs, and then run it. Maybe upload only one file. You can see the exact Internet communication in the FTPLOG files. You can also send them to support.

[tobias]

Hello,
you should also check bandwidth usage in Task Manager to check if the downloads are really done by Syncovery.

And maybe your Graphs are misleading. Syncovery receives confirmation for each uploaded chunk. But that's just a few Bytes.

[Jean-Michel]

Thanks
I will try with 1 file and turn-on the logging as you suggested