Best Practices for SGeRAC and Oracle RAC on HP-UX 11i, March 2009
10
Alternate configuration – fast reconfiguration with low heartbeat timeout
Figure 4. Alternative configuration – low Serviceguard member timeout
When RAC-DB-IC traffic is very high and SG-HB timeout is low, there is a probability of RAC-DB-IC
traffic interfering with SG-HB traffic and causing unnecessary timeouts. If SG-HB timeout can not be
increased, then an alternative action is to use a second network for SG-HB. This configuration is for
environments that need fast failover (low Serviceguard member timeout) for two or more nodes.
Each primary and standby pair protects against single failure. With SG-HB on more than one subnet,
a single subnet failure will not trigger a SG reconfiguration. If the subnet with CSS-HB fails, unless
subnet monitoring is used, CSS will resolve the subnet interconnect failure with a CSS cluster
reconfiguration.
Note:
Starting with Serviceguard A.11.19, the faster failover capability is
integrated with the base Serviceguard product. This configuration can be
used for faster failover.
Timeouts
The SG-HB timeout (MEMBER_TIMEOUT) should be set based on service availability requirements.
Optionally if CSS-HB timeout is changed, the CSS-HB timeout should be tuned to provide an
opportunity for Serviceguard to complete reconfiguration and update CSS through group membership
service (GMS) prior to CSS timeout. Optionally if RAC-DB-IC timeout is changed, RAC-DB-IC timeout
should be 15 seconds above CSS-HB timeout. Note that the timeout relations are slightly different
when using Cluster Interconnect Subnet monitoring.
Subnet monitoring or Cluster Interconnect Subnet monitoring
Serviceguard packages can be configured with a subnet dependency on the CSS-HB subnet and
FAIL_FAST enabled. If both LAN1 and lan2 failed, the Serviceguard package can request halting the
node where the interconnect failure is detected. Use of Serviceguard subnet monitoring has a
Node A
LAN 1
Node B
SG
-
HB2
Private
(primary)
Private
(standby)
Private
(primary)
LAN 1
LAN 2
SG
-
HB1
CSS-HB
RAC-DB1-IC
Private
(standby)
LAN 2
LAN 4
LAN 4
LAN 3
LAN 3