What an amazing product!
I manage a repository of software installers for volunteers who work in language development around the world. The internet in the villages where they live is often unreliable, expensive or non-existent. When these people go to town they can use a script that I developed that runs rsync to connect to our server and download any installers that have been updated since last time they connected. Back in the village, without any need for internet connection, they can install updates they have downloaded during the short time they were in town.
There is also a need for an always-on sync for some of these users, so that whenever they connect to the internet, any updated installers will sync to their computers automatically. When bittorrent sync was free we used that, but it now connects to Resilio sync, which is only free for individuals. It would be pretty expensive for the number of users who connect to our repository. SyncThing would do the job, but I would need to grant permission to each user for each tree of files they want to get.
I'm looking for something people could use to connect to our server, specify which trees of files they want and then forget it. Whenever they connect to the internet they'll get updates. We already run the rsync server -- would Rclone on our clients' machines do what I want?
Here's what I have read about Rclone already:
It is based on rsync.
The Ubuntu man page pointed me to rclone sync , where I found rclone sync -i SOURCE remote:DESTINATION
Could I turn that around and let users run rclone sync remote:SOURCE DESTINATION
If so, would this command go to the background and keep running, or would it need to be run each time a user wants to update?
Perhaps I could run something like this on our server:
rclone rcd --rc-web-gui
and people could run Rclone and set up a sync relationship where they use the web interface to specify which trees of files they want to sync.
The script that people can use at present, based on rsync, comes with a config file they can modify to uncomment the names of folder trees that they want to get. Something like that would also be suitable if I have misunderstood what --rc-web-gui provides.
Thanks for thinking about these things, and looking forward to your guidance -- Rclone can be used in so many ways and has so many options that I'm sure your experience will save me a lot of time.
I recommend you consider rsync and rclone as completely different tools.
They have some common functionality like synchronizing files to a remote server, but once you dig a bit deeper in the possibilities and limitations, they are quite different.
One important difference in your context is that rsync can do delta transfers (only changed parts of a file are transferred) while rclone always transfers the complete file.
Yes, you can sync in any direction and also between remote storages.
The sync would need to be started (manually or by a script).
I cannot follow you here, it seems like you got some rclone concepts mixed.
Tip: I learned rclone by watching a few videos on basic usage and then played with simple examples relevant for my for my use case while searching/reading the docs.
Idea: You may also want to consider an off the shelf cloud storage solution like Dropbox, Google Drive etc. Some of them have options to share files between accounts and then users can select which folders/files to keep permanently on the computer and performs file delta downloads - and they auto sync when connected to the internet (on non-metered connections).
I think this might be better idea specially with spotty and slow internet access. Dropbox is particularly good in delta syncs. It depends on the amount of information but a free account can host 2GB (easy to increase that limit if needed, or grab a premium account and share between all of them). They can upload/sync to Dropbox account and you could use rclone on your server to grab the content and sync them to your own server/service by using a polling sync command if needed.
Thanks, @Ole and @jcastill for your responses.
We had originally thought of using DropBox years ago, but someone raised the problem of a stolen computer being used to sync disguised keyloggers etc to our server. I don't think one-way sync was an option at Dropbox then. Perhaps it is now. I'll look into that.
Currently I am working on setting up a SeaFile server on our server and that might meet our need.
Thanks again for your answers. Rclone is not the solution I'm looking for, but I like the look of backing up to rsync.net for my own computer, and I would use Rclone for that.