Khepri DB path in Khepri enabled RabbitMQ cluster

[Vishnu Vardhan]

Hi Team,

I am trying to understand the difference in the metadata_store path after khepri is enabled.

In the mnesia enabled RabbitMQ cluster "RABBITMQ_MNESIA_BASE" will decide where to store its node metadata etc.

But once the Khepri_db feature flag is enabled and logs showing that migration from mnesia to khepri is finished according to - (https://www.rabbitmq.com/docs/metadata-store/how-to-enable-khepri#migration)  I dont see ant difference in the above mnesia base path

So can you please clarify the below queries

1. The above documentation says that mnesia records will be copied and then deleted, so what is the destination path for this copy, will a directory called khepri will be created ?
So that I can understand that this is where entire node data will be stored going forward

2. Apart from the applications logs showing that khepri is enabled, is there any verification that application is actually using khepri and not mnesia.

[Michal Kuratczyk]

To verify khepri is used, use `rabbitmqctl list_feature_flags` and check kepri_db's flag status.

You can also run `rabbitmq-diagnostics metadata_store_status`. It'll provide Khepri cluster details if Khperi is enabled

or just answer "mnesia is used" if Khepri is disabled.

As for the paths - they don't change just because Khepri is enabled. The paths and the metadata store used are completely separate things.

The "mnesia" folder is a historical artifact we might clean up at some point, but it's risky, since a node startup would need to move files

around, which may lead to all kinds of issues. Quite likely the "mnesia" part will disappear from new clusters in the future, but will remain

for clusters upgraded from older versions. Just speculation though, we don't have any detailed plans for addressing this.

Somewhere (based on your installation method and settings) within your data folder. there's a "coordination" folder.

That's where Khepri's data reside.

--
You received this message because you are subscribed to the Google Groups "rabbitmq-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to rabbitmq-user...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/rabbitmq-users/b199eb08-30c2-4903-b7ab-e47e9d2ace5bn%40googlegroups.com.

--

Michal

RabbitMQ Team

[Vishnu Vardhan]

Thanks for the response

Few follow-up questions here 

1. Is it still a valid for me to set the value of RABBITMQ_MNESIA_BASE even though I am using Khepri?

2. If the answer for above question is "yes", then according to the description of RABBITMQ_MNESIA_BASE will the node's database, message store, cluster state files etc. continue exists in the RABBITMQ_MNESIA_BASE path with Khepri also?

The major use case for me here is backup of message store, metadata etc. till now with mnesia I was taking the backup of files, directories at path "RABBITMQ_MNESIA_BASE" , so with khepri introduction does the same logic work or any extra files to be considered like you mentioned that mnesia part will be removed in future

[Vishnu Vardhan]

Thanks for the response

Few follow-up questions here

1. Is it still valid for me to set the value for RABBITMQ_MNESIA_BASE with khepri enabled ?

2. If answer for above question is "yes", as per the description of RABBITMQ_MNESIA_BASE will the message store, metadata etc. continue to get stored at path RABBITMQ_MNESIA_BASE with khepri introduction

The major use case for me with RABBITMQ_MNESIA_BASE is backup of message store, metadata etc. So even with khepri does the logic of taking backup of files at path RABBITMQ_MNESIA_BASE will still be valid?

On Monday, September 29, 2025 at 12:57:31 PM UTC+5:30 Michal Kuratczyk wrote: