How to deploy replication instance on different machines?

[kotaho]

Referred to : https://memgraph.com/docs/configuration/replication
How to deploy replication instances on multiple physical machines?
According to the docs given by memgraph, it seems that currently only deploying multiple instances on one machine is available, but this resulting in a very limit amount of available memory for per instance, especially in huge data senario of billions of nodes or edges, this could be a problem.

[antejavor]

Hi @kotaho, thanks for asking this question.

The docs contain just a boilerplate for setting up the replication.

The replication is supposed to host instances on different machines.

That means that you need to change the IP or DNS from the 127.0.0.1 or local host to something more appropriate in your case.

Do you see a way we could improve docs in that sense?

[kotaho]

Hi @kotaho, thanks for asking this question.

The docs contain just a boilerplate for setting up the replication.

The replication is supposed to host instances on different machines.

That means that you need to change the IP or DNS from the 127.0.0.1 or local host to something more appropriate in your case.

Do you see a way we could improve docs in that sense?

Thanks @antejavor
It would be very kind of you presenting some more detailed steps for settings of replication instances on different machines, or in terms of which the docs is improved
Look forward to suggestions from you

[antejavor]

@kotaho we are in the process of improving our docs for deployment. This will get the spot on the to-do and priority; we will do it in the following weeks; thanks for the feedback. 🙇‍♂️

[antejavor]

Hi @kotaho, what ports were exposed during the replication setup of the Docker?

[antejavor]

The probable issue here is that we didn't document that you need to open ports that are used in the replication process for replicas and main (port 10000 by default). This has to be added to docs when setting up replication with docker.

Locally, it will work because of docker networking (since containers are in the same network); in a real setup, it won't work.

We will add these changes to stage 2. of deployment best practices.

cc @kgolubic @katarinasupe