Bug or obscure info page with --dedupe-mode interactive

What is the problem you are having with rclone?

What is your rclone version

rclone version
rclone v1.52.3

  • os/arch: windows/amd64
  • go version: go1.14.7

Which OS you are using and how many bits

W10PRO x64

Which cloud storage system are you using?

Drive

The command you were trying to run

rclone --config=_config.txt --dedupe-mode interactive dedupe remote:

The rclone config contents with secrets removed.

not needed...

A log from the command with the -vv flag

in full glory in full glory

after stripping pacing and errors

2020/08/20 17:48:35 DEBUG : rclone: Version "v1.52.3" starting with parameters ["rclone.exe" "--config=_rclone-conf.txt" "-vv" "--dedupe-mode" "interactive" "dedupe" "remote:"]

2020/08/20 17:48:35 DEBUG : Using config file from "_rclone-conf.txt"

2020/08/20 17:48:35 DEBUG : Google drive root '': root_folder_id = "0ACFROvxSiJp0Uk9PVA" - save this in the config to speed up startup

2020/08/20 17:48:35 INFO : Google drive root '': Looking for duplicates using interactive mode.

===========================================================
quite a lot of "pacer" "low level retry" "User Rate Limit Exceeded"
finally at the end
===========================================================

2020/08/20 17:50:41 NOTICE: _GCC/^_fpLvgTO9h69k-I8B_7-SP2xB/^_jO5gBseFv9gTYplRXtKDAU1F24G-bcwltUd_7R4iBF03: Found 2 duplicates - deleting identical copies

2020/08/20 17:50:41 NOTICE: _GCC/^_fpLvgTO9h69k-I8B_7-SP2xB/^_jO5gBseFv9gTYplRXtKDAU1F24G-bcwltUd_7R4iBF03: Deleting 1/2 identical duplicates (MD5 "a60a5a38572478878653c703195c474d")

2020/08/20 17:50:42 INFO : _GCC/^_fpLvgTO9h69k-I8B_7-SP2xB/^_jO5gBseFv9gTYplRXtKDAU1F24G-bcwltUd_7R4iBF03: Deleted

2020/08/20 17:50:42 NOTICE: _GCC/^_fpLvgTO9h69k-I8B_7-SP2xB/^_jO5gBseFv9gTYplRXtKDAU1F24G-bcwltUd_7R4iBF03: All duplicates removed

Note that folder name and file names are encoded!

========================================================

So, I was expecting and interactive part... is the default behaviour, I put --dedupe-mode interactive in full purpose!!!

rclone just deleted dups without asking... Where was the interactive part?

I believe the run was ok and ther is no bug.... but my understanding wrong... the essential part is that of " The dedupe command will delete all but one of any identical (same md5sum) files it finds without confirmation. This means that for most duplicated files the dedupe command will not be interactive."

I believe the info page is obscure in wording, as speaks about files with same name and files with same name and contents. All to be called duplicates but only some of them really "identical" (clones ;--) The mix is a little ackward.

I really believe the info page needs some enlightenment!

========================================================

Then, about the info page for rclone dedupe, proposal to uplift text (if my proposal is not reflecting reallity, then the issue could be deeper...)

rclone dedupe

Removal of duplicated folder names inside same folder and file names inside same folder.

Synopsis

In the first pass it will merge directories with the same name iinside the same folder, traversing the full tree, following the tree branch, trunk to leave, then to next branch.

In the second pass, searches for files with duplicate names inside the same folder and resolves the duplication. Can be interactively (offering the user the way to resolve -default behavior) or non interactively (following the user specified behavior). Note that if there are files with same name inside the same folder that have same contents (md5) all but one will be deleted automatically even in interactive mode.

=== keep all from "Here is an example run." until "Dedupe can be run non interactively using the --dedupe-mode flag or by using an extra parameter with the same value", but not retain this "Dedupe can be run non interactively using the --dedupe-mode flag or by using an extra parameter with the same value", replace it by:

Dedupe can be run interactively or non-interactively. Interactively is the default behaviour or the behaviour followed if used the with command and parameter "--dedupe-mode interactive" or just the parameter "interactive". The non-interactive behaviour is achieved by using the command & parameter as follows - or simply the parameter"

I am not sure I did understood well the parameters implications... Proposal:

skip
says - "removes identical files then skips anything left"
questions
By design, same name same contents, all but one, are deleted. So, there are not identical files any more... but can be files with same name...
Is this implying
skip = "remove all but one of the identical (name+contents) files, skip the rest of files with same name different content"
or could be
skip = "remove all of the identical (name+contents) files, skip the rest of files with same name different content"

if so...
first = "remove all but one of the identical (name+contents) files, then keep the first of the same name different content."
or could be
first = "remove all of the identical (name+contents) files, then keep the first of the same name different content."
or could be
first = "keep only the first of the same name."

newest/oldest/largest/smaller (change x first)

rename
says - "removes identical files then renames the rest to be different."

rename = "remove all but one of the identical (name+contents) files, then rename the rest of same name files to different names"
or
rename = "remove all of the identical (name+contents) files, then rename the rest of same name files to different names"

TIA

You are correct, rclone will delete identical files where size+name+hash is identical without asking even with --dedupe-mode interactive

Perhaps we should have another mode semi-automatic which works like it does now and make interactive query every deletion.

I agree this is all a bit confusing - maybe rclone should just not do that auto deletion phase.

BTW I think I changed the wording in the latest beta so you might want to take a look at that

This topic was automatically closed 60 days after the last reply. New replies are no longer allowed.