Seafile synchronization after downtime


#1

I’d like to understand a specific case of synchronization. We use seafile in our company and I manage all the IT activities related to the service. Some days ago we had to update our ubuntu server to 18.04 and we had to do a clean installation. Unfortunately we realised our db dump was 12 hours older than the data (for some errors in our procedure). Long story short, the whole update took us 1 entire day and in the meantime people were working on their local folders offline with the hope that the changes could have been updated when the service was back online. Instead we realised that when back online, the synchronization was towards the clients, so all the changes made offline were overwritten by the old copy. (we tested this before restarting the service for everyone and we were able to warn all the people)
My question is then, why is this happening? should be the most recent copy the one seafile should keep?
thanks for your help


#2

It’s really scary your experience.


#3

Thank you for empathizing jacky :smiley: now we know some more things.
I think we can and should improve a lot our procedure, but the last aspect, regarding the synchronization, was quite surprising for us. I’d like to understand if we missed something in that part, if it is really supposed to work that way, if we could tell seafile how to behave. Any suggestion is welcome


#4

In my opinion this is a bug in seafile-client which needs to be fixed. The very minimum should be the generation of conflict files.

@daniel.pan


#5

I thought so, I agree, but I’m not sure about the conflict copy: you would have them on shared folders as people is basically working offline and it might happen they touch the same file, so when you’re back online you have a conflict, but on “clean” files I expect to just keep the most recents when the service goes up again. So what people has, not what the seafile server has.
By the way, did you have a similar experience?


#6

I didn’t have any issues so far but also didn’t have to recover my backup in the past years and it’s only a personally used server.

I keep using Seafile because the synchronization works quite well for about 5 years now and I did have massive issues with other solutions.

Unfortunately apart from the synchronization my satisfaction with Seafile goes down year by year. I’m neither satisfied by the update process and quality control nor with the development it underwent in some areas.

E.g. the implementation of the Markdown Editor was waste of resources in my opinion and also added tons of Javascript from tons of sources. I also miss a well implemented photo viewer with precomputation of thumbnails using a background service. Currently viewing photos is a pain and large thumbnails even seem to be baseline jpegs (instead of progressive) also the server will be overloaded quite fast when opening a folder because multiple thumbnails need to be generated for each file and one has to wait for them. The filesearch also still is not really good and a basic filename / path search is still missing (I think the search has been improved recently but I haven’t tried it out so far - I wait for another bugfix release before I update - perhaps).

Also SeaDrive which was a good idea still has many major flaws.