Hi,

On Mon, Jan 13, 2020 at 1:19 AM Marcel Reutegger <[email protected]>
wrote:

> Hi,
>
> On 12.01.20, 15:40, "jorgeeflorez ." <[email protected]>
> wrote:
>
> > If I create two backends
> > (Oak instances), both using the same type of document and blob store,
> > and both pointing to the same "location" (folder in a file system, S3
> > path, etc). will they work without collisions or conflicts when
> > reading/storing files?
>
> For the blob stores this is generally true. When it comes to NodeStore
> implementations, only the DocumentNodeStore will also work in a clustered
> setup. The SegmentNodeStore implementation does support multiple active
> processes working on the same storage.
>

I'm pretty sure the blob stores need to implement SharedDataStore to work
in this context - or at least if you want data store garbage collection to
work in these data stores.  I'd use OakFileDataStore if you want to use a
filesystem-based blob store, S3DataStore for S3 and AzureDataStore for
Azure - they all implement SharedDataStore.


-MR

Reply via email to