What counts as uploaded?

[Christoph]

Just a quick question regarding the statistics for the backup command: what exactly counts as uploaded. More specifically:

2018-10-10 15:10:41.059 INFO UPLOAD_FILE Uploaded Users/Christoph/Dropbox/myscript.ps1 (1443)

Given the possibility that this file might already have been backed up by another machine (synced via dropbox), does this message mean that the file was previously not present in the storage and the relevant chunks have now been added to the storage? Or is it possible that the chunks were already in the storage but have now been added to this particular repository snapshot?

Accordingly here:

2018-10-10 15:10:41.347 INFO BACKUP_STATS File chunks: 47036 total, 54,252M bytes; 143 new, 185,490K bytes, 95,082K bytes uploaded

Does this mean that 95,082 kB of chunks were actually uploaded or does this figure include skipped junks?

[gchen]

Or is it possible that the chunks were already in the storage but have now been added to this particular repository snapshot?

That is correct.

P_STATS File chunks: 47036 total, 54,252M bytes; 143 new, 185,490K bytes, 95,082K bytes uploaded

Does this mean that 95,082 kB of chunks were actually uploaded or does this figure include skipped junks?

143 new chunks with a total size of 185,490K were uploaded. The actual upload size is 95,082K due to compression. Skipped chunks are not counted here.

[Christoph]

Or is it possible that the chunks were already in the storage but have now been added to this particular repository snapshot?

That is correct.

In that case I suggest to change the message to “Backed up” instead of “Uploaded”. Whether the technical term UPLOAD_FILE should be changed accordingly is a matter of taste. I would prefer it for the sake of consistency.

[mfeit_duplicacy]

“Backed up” instead of “Uploaded”

Uploaded is technically correct in this context because by that part of the software, it’s all chunks sent to storage. Duplicacy happens to make note of how many originated as files and how many originated as metadata.

There’s already information in the output that covers the number and size of files in the backup set and how much of it changed:

     Backup for /foo/bar/baz at revision 355 completed
-->  Files: 1251173 total, 1443G bytes; 2559 new, 6,449M bytes
     File chunks: 304798 total, 1452G bytes; 799 new, 3,989M bytes, 3,404M bytes uploaded
     Metadata chunks: 114 total, 486,492K bytes; 37 new, 155,823K bytes, 50,012K bytes uploaded
     All chunks: 304912 total, 1452G bytes; 836 new, 4,141M bytes, 3,453M bytes uploaded
     Total running time: 00:14:35

[Christoph]

Uploaded is technically correct in this context because by that part of the software, it’s all chunks sent to storage.

I don’t want to be splitting hairs, but if “upload” means “to transfer information to the storage”, then this is not correct usage of the term because the information has been not been transferred to the storage (it was already there). Instead of uploading it, a reference to that existing information has been created in the snapshot file.

[Christoph]

Okay, just to clarify @gchen, I’m suggesting that duplicacy be more precise in it’s language and use the term “uploaded” only at least one chunk has actually been transferred and “backed up” when the file has merely been added to the snapshot but not chunks needed to be uploaded.