===== Update from Jewel to Luminous =====

Accertarsi che tutti i MON e tutti gli OSD siano up. Accertarsi, con "ceph status', chelo stato sia HEALTH_OK.

Poi da ceph-mon-01:

<code bash>
ceph osd set sortbitwise
</code>


Poi sempre da ceph-mon-01:

<code bash>
ceph osd set noout
</code>


==== Update MON ====

Su ceph-mon-01, poi ceph-mon-02 e infine ceph-mon-03:

<code bash>
rpm -e ceph-release
rpm -Uvh https://download.ceph.com/rpm-luminous/el7/noarch/ceph-release-1-1.el7.noarch.rpm
yum clean all
yum update

shutdown -r now
</code>

Terminato l'update dei monitor verificare da un mon le versioni con:

<code bash>
ceph versions
ceph features
</code>


Da un mon:

<code bash>
ceph mon feature ls
</code>


==== Configurazione manager ====

Da uno dei 3 mon, creare le chiavi per i manager:

<code bash>
ceph auth get-or-create mgr.ceph-mon-01 mon 'allow profile mgr' osd 'allow *' mds 'allow *'
ceph auth get-or-create mgr.ceph-mon-02 mon 'allow profile mgr' osd 'allow *' mds 'allow *'
ceph auth get-or-create mgr.ceph-mon-02 mon 'allow profile mgr' osd 'allow *' mds 'allow *'
</code>

Su ceph-mon-01:

<code bash> 
mkdir /var/lib/ceph/mgr/ceph-ceph-mon-01
chown ceph.ceph /var/lib/ceph/mgr/ceph-ceph-mon-01
</code>

Creare il file /var/lib/ceph/mgr/ceph-ceph-mon-01/keyring

che deve avere questo ownership/mode:

<code>
-rw------- 1 ceph ceph
</code>

e un contenuto di questo tipo:

<code bash>
# cat /var/lib/ceph/mgr/ceph-ceph-mon-01/keyring
[mgr.ceph-mon-01]
	key = AQBkuC5aQMCyHhAASi2BbZghbzG7f6vcHXPS8Q==
</code>

con il valore della chiave che deve matchare quanto ritornato da "ceph auth get":

<code bash>

# ceph auth get mgr.ceph-mon-01
exported keyring for mgr.c-mon-1
[mgr.c-mon-1]
	key = AQBkuC5aQMCyHhAASi2BbZghbzG7f6vcHXPS8Q==
	caps mds = "allow *"
	caps mon = "allow profile mgr"
	caps osd = "allow *"
</code>


Abiltare e far partire mgr:

<code bash>
systemctl restart ceph-mgr@ceph-mon-01.service
systemctl enable ceph-mgr@ceph-mon-01.service
</code>


Verificare con "ceph status" che il monitor sia attivo

Fare le stesse operazioni sugli altri 2 monitor node.


==== Update OSD ====

Su ceph-osd-01, poi ceph-osd-02 e infine ceph-osd-03:

<code bash>
rpm -e ceph-release
rpm -Uvh https://download.ceph.com/rpm-luminous/el7/noarch/ceph-release-1-1.el7.noarch.rpm
yum clean all
yum update

shutdown -r now
</code>

Terminato l'update degli OSD verificare da un mon le versioni con:


<code bash>
ceph versions
ceph features
ceph osd versions
</code>

==== Operazioni da fare prima di updatare i client ====

Da un minitor:

<code bash>
ceph osd require-osd-release luminous
ceph osd unset noout
</code>

In caso di warning (con ceph health) del tipo "HEALTH_WARN application not enabled on 3 pool(s)" taggare i pool in questione, e.g.:

<code bash> 
ceph osd pool application enable pool1 test-pool
</code>

==== Attivazione della dashboard ====

Da un monitor:

<code bash>
ceph mgr module ls
ceph mgr module enable dashboard
</code>

Verificare che la dashboard sia stata abilitata e trovare la relativa URL:

<code bash>
ceph mgr module ls
ceph mgr services
</code>

==== Update dei client ====

Su ogni client:

<code bash>
rpm -e ceph-release
rpm -Uvh https://download.ceph.com/rpm-luminous/el7/noarch/ceph-release-1-1.el7.noarch.rpm
yum clean all
yum update

shutdown -r now
</code>

Terminato l'update, verificare da un mon le versioni con:


<code bash>
ceph features
</code>