about ctdb deployment without clustered file system

[Martin Schwenke]

Hi,

On Sat, 15 Feb 2020 12:08:08 +0800 (CST), 风无名 via samba-technical
<samba-technical at lists.samba.org> wrote:

> evryone.I'm sorry to bother you guys. I want to ask  a question about ctdb.
> 
> In ordinary config, we need a clustered file syste, such as
> glusterfs/ceph to provide a sharing volume for ctdb.
> 
> If I do not deploy any clutered file system,  use
> samba/ctdb/utils/etcd/ctdb_etcd_lock (maybe it needs
> rewritten/modified)  instead of the file lock, and place every config
> file in local servers instead of the sharing volume, will ctdb work
> well ?
> 
> I have tested some situations and not found anything abnormal.I am
> afraid that the tests are not enough.

CTDB will work in this mode, as a replicated and distributed database
without any problems.

However, CTDB's main job is to provide a distributed (and, depending on
the database, replicated) database for Samba file metadata. If you have
smbd instances running on multiple nodes but do not have a cluster
filesystem then the different smbd instances can not service the same
files.  This means you don't get either the scale-out offered by
Clustered Samba or the redundancy.

So it will work, but I wonder why you would do it...  :-)

peace & happiness,
martin