Please add a tick box just beside the button “override change” to always override the change.
I’m working on a network with only low speed connections between hosts so I want to keep the “trusting” between hosts but, The new data is always coming from one hosts “server-like”, and populated over the others so I need a ticker on the “Master-repo” to always override changes over all the network to ensure all the other hosts will be in sync.
No. This has been discussed to tediousness here already. To prevent changes on your “slaves”, use filesystem permissions as appropriate there. Allowing changes and then immediately undoing them is wasteful, confusing and potentially dangerous.
… couldn’t this be a second option beside “master device” that is disabled by default? (maybe in advanced options)
I think the idea is charming - because clicking “override” a thousand times s***
This is the root cause and what we should be solving instead. Tell us more about these machines. OS, version, filesystem? Also, what kind of changes is the master claiming to see / not be up to date with?
Note that this could be something simple like an auto generated file. For example, if those machines are Macs, they’ll be creating .DS_Store files in directories whenever a user customizes the view of it in the Finder. Syncthing will announce these, and the master becomes out of sync. If you press “override” they’ll be deleted, and the cycle starts over. Ignoring these on the client is the best bet, but https://github.com/syncthing/syncthing/issues/1631 would be nice to have solved to have directory deletes work better in that environment… You can also ignore the fact that it says “out of sync” if it’s these kind of changes that cause it - it does no harm.
The reason why is on the forums search for it. If there is suddenly two masters, they would keep overwriting each others changes forever. Also, it could potentially nuke your data, if for example you try editing a file on a non-master.
If you want the files to be read only, sync them as read only, or don’t modify them after receiving, rather than writing ahk scripts.
i have the same problem with the override feature.
i am using syncthing to replicate a Master Repo and some of the guys out (service techs) are doing nonsens out there. unpacking files for example. but they are unpacking these into their local sync-directory.
however, these files are replicated to all other users, because of the flag to push files also to other clients. it is wanted to have this feature, due to bandwith optimisation.
but: i cannot be online all the time to overwrite (and delete) the obsolete files on their node by pressing “overwrite changes” always.
The Master Node is read-only. and i want to have this feature at the Master node. I know the risks… but in this case, it is really necessary…
What you are asking for is not bidirectional sync which is the point of Syncthing. I honestly think you could use rsync or similar but there are other options.
There are conversations of a slave mode that would suit your needs but is not implemented yet. You can achieve the same thing by having syncthing run as a different user and only allow the syncthing user write access to the folder you do not want the technicians writing into.
Rsync for 110 dynamic IP Users and a repo with 800GB over DSL and LTE via Internet and VPN ?
You are right, it is a read only Repo. But why there is an option for Read Only , if this is not an allowed option ?
I do not care about , where i can set the option , client or server.
but the clients should be able to sync the files between each other also.
Next step will be 3 master repos. Syncing via different Folder-ID´s than the clients, just to override the read-only flag
@AudriusButkevicius: The Servers would not get the Folder-ID to sync between each other with that always override flag. So they wont mess up.
So it is still bi-directional but also uni-directional.
server <-> server
client <-> client
but:
server → client
i am not sure, if rsync would be capable of this. but the comment of @calmh is good:
Master and Slave nodes…lookinbg forward to test this
so lets close this topic for now.
On BTSync it was possible, but i have to have a private cloud. And also Syncthing is OpenSource. I love this Project , so i have got my own DiscoServer and RelayServer (not yet completed)
You could do automatic override via the rest api and a cronjob, there is /rest/db/override which takes folder as argument from what it looks like at the code.
This is probably undocumented for reasons mentioned above (two masters always doing it or unexpected nuking of data on non-master), use at your own risk (or it was just forgotten to document it^^)
I can see the usefulnes of a known “slave” mode for all slave-devices: They would not accept local changes in the index, thus not removing / modifying data on other nodes, marking dirty blocks as “needed” for themselves, and still alowing peer-to-peer filetransfer of any unmodified data, even in case master is offline or connected very slowly.
I understand the reasons for not having this option, but what would really help in identifying the problem: A way to display the differences. The red button really doesn’t help there…