Syncing two collections, on different installations of Appwrite

Hey do you have any code sample for this?

Is the problem that server A has data and it gets synced to B, but then B tries to sync that back to A creating a loop?

Huge code right now, because I'm testing lot of things.

Fair enough

Think like, server B (remote) gets new data, a function catch that data with an event, and put it on a collection called SYNCS, that collection is a queue to sync to server A (local).

On server A there is a function that set flag ON in server B (remote), so server B will ignore new events (during the execution of this function).

Then, the same function starts syncing data on queue (SYNCS collection) the problem, and finish the process setting the flag OFF.

Problem: The Function A (the one that is triggered by an event) is executed after Function B finishes, so FLAG is OFF by that time, so it will generate new SYNCS documents, that is not the intended.

I could try to make a diagram if you guys need to understand better

let's take a step back and don't think about your implementation...what are you trying to do?

That's a good Idea.

I have two appwrite instances, one running locally, and another on internet (call it remote).

In both, there is a collection, and I want to sync bidirectionally this collections. Like, if I add something to the collection on the remote server, it will replicate the same on local. The same happens in local, if I add something on its collection, it will replicate that on remote.

This have to work for CREATE, UPDATE, DELETE

And so if a document is synced from remote to local, you don't want that change to get synced again from local back to server in an infinite loop, right?

Yes sir!

Right so you need a way to know there's no need to update...

I think you're going to have to store something on the document to know. Maybe you can store who updated the document: sensor or sync. If it was updated by sync, don't sync

I though on the same, so I will implement that, to check how it works with that

but actually, reading from you that work around points me in the right direction

thank you so much Steven

And sorry for my late reply, working a lot today!

no worries!

@Ponguta_ everything good? Checking to see if it makes sense to close this and mark as solved 🙂

It's needed to mark it as solved?

Checking if everything is good and if you still needed any assistance

Just as a little update on this.

IDK why this happens, if is just me, or what. buuut.

Seems like the time passed from the request made with databases.update_document('[DATABASE_ID]', '[COLLECTION_ID]', '[DOCUMENT_ID]') and the execution of a function that catch that event is around 6.5 - 7.5 seconds.

What I mean with this.

If I set the a flag ON, to stop catching events, this flag should be ON for at least that amount of time, before going OFF, otherwise, it will be useless.

events get queued and, on self hosted, they're executed serially. So, if you have other events going on, they may be processed first before this update document one is processed

any way to do it faster? or thats just how it works

You would have to scale out horizontally with more workers that can handle the events

Any guide on how can I do that?

I really want to try that out Steven

See https://discord.com/channels/564160730845151244/1096317501404610570