G A L A X Y N A S L X 5 C O N F I G U R A T I O N A N D S Y S T E M I N T E G R A T I O N G U I D E
48
Section 3 Menu Functions
performed by the passive node. If this one succeeds, the nodes perform the failover. The ping node
needs only to be able to receive and reply to ICMP echo requests.
Serial port:
This options allows you to connect the node via a serial cable. Select the port to which the given cable
has been connected.
Keep failover functionality after volume replication failure:
When activated, this option will allow automatic failover to be carried out after the volume replication
between the nodes has been broken. Please note that in certain circumstances this may lead to data
loss. Details can be found in the pop-up window which appears after ticking the checkbox for this
option.
Note:
Warn, dead and init timers are started simultaneously.
Note: Default values for timers:
Warn time - 5000ms
Dead time - 10000ms
Init time - 15000ms
Keep alive time - 1000ms
Note: Configuring the cluster with low warn, dead and init timers may result in
unstable behavior. Optimal timer settings are dependent on server hardware and performance.
Note: Before starting failover please make sure the LUN configuration is the same on
both nodes, i.e. that the LUNs are of the same size and type (block I/O or file I/O) and
that the LUN
order is preserved.
WARNING: Inconsistencies in this area may lead to data loss!
Note:
If you see the following error message: Service is under heavy load or the
time values you have set are too low, you need to readjust the time values (they are
dependent on your network connectivity and system load). When using low deadtine
value,
one of your nodes may wrongly assume the other one is dead, which will lead to a
situation
where both nodes are active (the so-called "split brain").
