HP Hitachi Dynamic Link Manager Software - Manual (Page 297)

3. Creating an HDLM Environment
273
•
rr
for the Round Robin algorithm
•
exrr
for the Extended Round Robin algorithm
•
lio
for the Least I/Os algorithm
•
exlio
for the Extended Least I/Os algorithm
•
lbk
for the Least Blocks. algorithm
•
exlbk
for the Extended Least Blocks algorithm
The type of algorithm specified by the
-lbtype
parameter remains stored in the
system, even if, you disable the load balancing function by specifying
-lb off
. If you
decide to re-enable load balancing at a later time, and you do not specify an algorithm,
load balancing will still be executed by using the algorithm that was previously
specified.
(2) Setting Up Path Health Checking
You can choose whether you want to use path health checking.
The following is an example of using a command to enable path health checking.
# /opt/DynamicLinkManager/bin/dlnkmgr set -pchk on -intvl 10
To enable path health checking, specify
on
. Specify
off
otherwise. When you specify
on
, you can use the
-intvl
parameter to specify the checking interval. The value that
was specified previously will be applied if the checking interval is not specified. For
example, specify the path health check as
off
after specifying the checking interval as
15
minutes and executing. Then, when executing after specifying the path health check
as
on
without specifying the checking interval, the
15
minutes that were specified
previously will be applied again.
(3) Setting Up the Automatic Failback Function
When intermittent error monitoring is enabled and the number of error occurrences is
2
or more, the following condition must be satisfied.
error-monitoring-interval >= checking-interval-for-automatic-failbacks x
number-of-times-an-error-is-to-occur-during-intermittent-error-monitoring
If this condition is not satisfied, an error will occur and the warning message
KAPL01080-W will be output.
If this happens, change any of the following settings: the checking interval for
automatic failbacks, the intermittent error-monitoring interval, or the number of times
that the error needs to occur.