Even aside from Galera, when you start a 10.1 server on a 10.0 datadir, it should produce lots of error messages regarding wrong structure of system tables and such. Did you have those?
Why would you want to run a Galera node (or any server, for that matter) on a non-upgraded datadir? Obviously it won't work properly.

^{I made an upgrade before (forgot to say)}

But you said specifically in the description "without update datadir" – what did it mean, then? Did you run mysql_upgrade on this server?

omg, sorry forgot my last comment i didn't upgrade (i ask myself how i could forgot that). I made so many test in same time

So, back to the question – what was the expected result of this experiment for you? Did you expect the new server would work on the old datadir? Or what exactly do you want to be fixed?

the problem is no explicit message. and the error long error tram.

The long log of error messages is unavoidable, because the server must start on the old datadir (or how else could you upgrade it), and all components that encounter problems must log errors about it.
I'll convert your report into a feature request for an explicit error message about a non-upgraded datadir.

start with SST but couldn't be good because not the good version of systemes table.

In fact I have to back from begining lol

nirbhay_c,

If not done yet, It would be good to document the recommended process of upgrading MGC nodes to a higher major server version.

I try to do that actually

Impossible to reproduce plz close it

i tried with new node

and I thought galera was include with MAriaDB and in Centos 6.x and i got one package for MariaDB server and one other for galera.

I tried to migrate and I made it without any problem.

stop server
remove package
update repo
install package
start server
and all fine except for some system table, but mysql started and joined cluster (first node was already in 10.1.6 and in cluster)