[Cluster] Indisponibilité VMDK

Bonjour,

Nous avons un plateforme à 4 nœuds (sous VMware) avec un volume monté en gfs2 sous le nom de « shared ».

Suite à plusieurs tests, nous avons contester que le volume (dossier nommé shared) n’est plus disponible lorsque qu’une machine virtuelle est arrêté.

Voici mon test que j'ai pu effectuer afin de vérifier la disponibilité du dossier partagé « shared »:

• Arrête complet de la VM1.
• Executer la commande depuis la VM2, VM3 et VM4 « ls /shared » afin de vérifier la disponibilité du dossier « shared ».

Résultat :

[user@VM2 ~]$ ls /shared/

[user@VM3~]$ ls /shared/

[user@VM4 ~]$ ls /shared/

Constat : une fois le serveur VM1 est éteint, le dossier « shared » n'est plus accessible depuis les autres VMs.
Une fois la VM1 démarré, le dossier « shared » est de nouveau disponible.

Voici la configuration cluster mis en place :

<?xml version="1.0"?>
<cluster config_version="45" name="cluster-nepal">
        <logging debug="on"/>
        <fence_daemon clean_start="1" post_fail_delay="0" post_join_delay="3"/>
        <dlm plock_ownership="1" plock_rate_limit="0"/>
        <gfs_controld plock_rate_limit="0"/>
        <cman broadcast="yes" expected_votes="3"/>
        <clusternodes>
                <clusternode name="nepal-vascep01-ach-val" nodeid="1" votes="1">
                        <fence>
                                <method name="1">
                                        <device name="fence_soap1"/>
                                </method>
                        </fence>
                </clusternode>
                <clusternode name="nepal-vascep02-ach-val" nodeid="2" votes="1">
                        <fence>
                                <method name="1">
                                        <device name="fence_soap2"/>
                                </method>
                        </fence>
                </clusternode>
                <clusternode name="nepal-dbag01-ach-val" nodeid="3" votes="1">
                        <fence>
                                <method name="1">
                                        <device name="fence_soap3"/>
                                </method>
                        </fence>
                </clusternode>
                <clusternode name="nepal-dbag02-ach-val" nodeid="4" votes="1">
                        <fence>
                                <method name="1">
                                        <device name="fence_soap4"/>
                                </method>
                        </fence>
                </clusternode>
        </clusternodes>
        <fencedevices>
                <fencedevice action="reboot" agent="fence_vmware_soap" ipaddr="10.153.37.17" login="fenceNepal@vsphere.local" name="fence_soap1" passwd="W@Fuste6ec" ssl="1" uuid="4228b4ce-fdc6-a323-24ff-5f0430824dc1"/>
                <fencedevice action="reboot" agent="fence_vmware_soap" ipaddr="10.153.37.17" login="fenceNepal@vsphere.local" name="fence_soap2" passwd="W@Fuste6ec" ssl="1" uuid="422858b9-b377-9855-db56-bfc8f96c3525"/>
                <fencedevice action="reboot" agent="fence_vmware_soap" ipaddr="10.153.37.17" login="fenceNepal@vsphere.local" name="fence_soap3" passwd="W@Fuste6ec" ssl="1" uuid="42284360-0e4b-531e-7e94-bb10505f1c74"/>
                <fencedevice action="reboot" agent="fence_vmware_soap" ipaddr="10.153.37.17" login="fenceNepal@vsphere.local" name="fence_soap4" passwd="W@Fuste6ec" ssl="1" uuid="42283948-ad7a-7e5a-5538-b34b01aa1f7d"/>
        </fencedevices>
        <rm log_level="7">
                <failoverdomains>
                        <failoverdomain name="cluster-vascep" nofailback="1" ordered="1" restricted="1">
                                <failoverdomainnode name="nepal-vascep01-ach-val" priority="1"/>
                                <failoverdomainnode name="nepal-vascep02-ach-val" priority="2"/>
                        </failoverdomain>
                        <failoverdomain name="cluster-dbag" nofailback="1" ordered="1" restricted="1">
                                <failoverdomainnode name="nepal-dbag01-ach-val" priority="1"/>
                                <failoverdomainnode name="nepal-dbag02-ach-val" priority="2"/>
                        </failoverdomain>
                        <failoverdomain name="vascep01" nofailback="0" ordered="0" restricted="1">
                                <failoverdomainnode name="nepal-vascep01-ach-val" priority="1"/>
                        </failoverdomain>
                        <failoverdomain name="vascep02" nofailback="0" ordered="0" restricted="1">
                                <failoverdomainnode name="nepal-vascep02-ach-val" priority="1"/>
                        </failoverdomain>
                        <failoverdomain name="dbag01" nofailback="0" ordered="0" restricted="1">
                                <failoverdomainnode name="nepal-dbag01-ach-val" priority="1"/>
                        </failoverdomain>
                        <failoverdomain name="dbag02" nofailback="0" ordered="0" restricted="1">
                                <failoverdomainnode name="nepal-dbag02-ach-val" priority="1"/>
                        </failoverdomain>
                </failoverdomains>
                <resources>
                        <ip address="10.153.37.15" monitor_link="1"/>
                        <ip address="10.153.37.16" monitor_link="1"/>
                        <ip address="10.160.60.116" monitor_link="1"/>
                        <ip address="10.160.60.244" monitor_link="1"/>
                        <script file="/etc/init.d/igloo" name="igloo-init"/>
                        <script file="/etc/init.d/cluster-vascep-check" name="vascep-check"/>
                        <script file="/etc/init.d/cluster-dbag-check" name="dbag-check"/>
                </resources>
                <service autostart="1" domain="vascep01" exclusive="0" max_restarts="3" name="igloo-init-vascep01" recovery="restart" restart_expire_time="0">
                        <script ref="igloo-init"/>
                </service>
                <service autostart="1" domain="vascep02" exclusive="0" max_restarts="3" name="igloo-init-vascep02" recovery="restart" restart_expire_time="0">
                        <script ref="igloo-init"/>
                </service>
                <service autostart="1" domain="dbag01" exclusive="0" max_restarts="3" name="igloo-init-dbag01" recovery="restart" restart_expire_time="0">
                        <script ref="igloo-init"/>
                </service>
                <service autostart="1" domain="dbag02" exclusive="0" max_restarts="3" name="igloo-init-dbag02" recovery="restart" restart_expire_time="0">
                        <script ref="igloo-init"/>
                </service>
                <service autostart="1" domain="cluster-vascep" exclusive="0" name="vip-vascep" recovery="relocate">
                        <ip ref="10.153.37.15">
                                <script ref="vascep-check"/>
                        </ip>
                </service>
                <service autostart="1" domain="cluster-dbag" exclusive="0" name="vip-dbag" recovery="relocate">
                        <ip ref="10.153.37.16">
                                <ip ref="10.160.60.116">
                                        <ip ref="10.160.60.244">
                                                <script ref="dbag-check"/>
                                        </ip>
                                </ip>
                        </ip>
                </service>
        </rm>
        <totem consensus="40000 + 5000" token="40000"/>
</cluster>

Bien cordialement,

Benjamin Soultany