How to get continuous upload?


I am wondering how to get a continuous upload stream. Now it uploads a block at full speed, and when that is finished a short wait and then another block. I have tried different sizes and amount of blocks, but no difference.

1 Like

Duplicati is threaded among the various processes processes that it does, however depending on individual machines, OSes, hardware, etc. it’s not uncommon for one of the threaded steps (say uploading a file) to finish before another of the threads (say zipping the next file to be uploaded) is done. This results in one of the threads sitting around waiting for the next piece of data in can process.

The specific thread-waits vary, for example:

  • if you have a slow CPU then the hash-lookups or calculations used for de-duplication could be the slowest step
  • if you have a slow hard disk then the indexing of the next set of files could be the slowest step
  • if you have slow internet then the uploading could tbe the slowest step

All that being said, certain features can be optimized or disabled. It’s tough to know what the specific things to adjust are for your scenario, but a few things you could try for testing purposes include going to Advanced Options and setting:

  • the asynchronous-upload-limit=# parameter can be used to set how many archive files to make “ahead of time”. I think the default is 4, so try increasing that value and see if you get a change in your “sawtooth” pattern
  • the check-filetime-only=true parameter can be used to make the change detection flow faster (at the cost of accuracy)
  • the dblock-size=50 parameter can be used to set the size of the archive files being created - making that value SMALLER should make the building of the archives a bit faster (at the cost of having more files)

If none of those have any effect then let us know as there are probably a few other parameters that could be adjusted to try and identify the bottleneck affecting you.

Oh - another thing you could try is keeping an eye on the Processes tab while duplicati is running. When the network usage is LOW (at which point we assume it’s just waiting for another file to be ready) what of CPU, Memory, or Disk gets HIGHER?

1 Like