TOPIC:

Hi there
I have set up a HALizard pool with an SMB storage.

If I switch off the master, the VMs are started on the slave, perfect.

But if I now e.g. disconnect the master from the LAN or switch port error occurs, the whole pool is deactivated, and I can't get anywhere. What do I have to change in the configuration, so the slave takes over and I still see the master in the Pool?

DISABLED_VAPPS=()
DISK_MONITOR=1
ENABLE_ALERTS=1
ENABLE_LOGGING=1
FENCE_ACTION=stop
FENCE_ENABLED=1
FENCE_FILE_LOC=/etc/ha-lizard/fence
FENCE_HA_ONFAIL=0
FENCE_HEURISTICS_IPS=8.8.8.8
FENCE_HOST_FORGET=0
FENCE_IPADDRESS=
FENCE_METHOD=POOL
FENCE_MIN_HOSTS=2
FENCE_PASSWD=
FENCE_QUORUM_REQUIRED=1
FENCE_REBOOT_LONE_HOST=0
FENCE_USE_IP_HEURISTICS=1
GLOBAL_VM_HA=1
HOST_SELECT_METHOD=0
MAIL_FROM="root@localhost"
MAIL_ON=1
MAIL_SUBJECT="SYSTEM_ALERT-FROM_HOST:$HOSTNAME"
MAIL_TO="root@localhost"
MGT_LINK_LOSS_TOLERANCE=5
MONITOR_DELAY=15
MONITOR_KILLALL=1
MONITOR_MAX_STARTS=20
MONITOR_SCANRATE=10
OP_MODE=2
PROMOTE_SLAVE=1
SLAVE_HA=1
SLAVE_VM_STAT=0
SMTP_PASS=""
SMTP_PORT="25"
SMTP_SERVER="127.0.0.1"
SMTP_USER=""
XAPI_COUNT=2
XAPI_DELAY=10
XC_FIELD_NAME='ha-lizard-enabled'
XE_TIMEOUT=10

Hi Heinz,
Below are some explanations of high level logic on what should happen. Ultimately, a log from the slave would be required to know what went wrong in your test case.

- If the master loses its MGT interface (link down), we detect that and shutdown all VMs running on the master after MGT_LINK_LOSS_TOLERANCE wait time has been exceeded. The slave is then expected to restart the VMs
- In order for the slave to take over services, it must be able to reach the configured heuristic_ip. In your case it looks like you have the default value set, which requires that the slave have an active connection to the internet to work. I suggest you change that to the local IP of the switch or nearest router.
- if both the master and slave cannot reach the heuristic IP AND cannot see each others MGT IP, then all VMs in the pool will shutdown. This is intentionally done to ensure that VMs cannot run on both hosts.

/var/log/user.log from the slave will provide some insight into why the slave failed to take over services.