[Linux-HA] Shortest monitor interval practical?
dejanmm at fastmail.fm
Thu Sep 20 04:13:39 MDT 2007
On Wed, Sep 19, 2007 at 12:19:00PM -0700, Joseph Lamoree wrote:
> Related to this, I would like to see some discussion of the
> relationship of interval to timeout. For example, in the cib.xml
> linked to the GuiGuide, the OCF:Filesystem resource monitor
> interval:timeout values are 120:60, while the LSB:nfsserver is 30:60.
> It would be helpful to know the reasons for these settings, since it
> might enable me to steer clear of issues down the road.
Both examples you quoted may make sense even though the
timeout/monitor interval ratio are very different. Doubt that
there was any deep philosophy behind those.
The most important thing to consider when dealing with intervals
and timeouts is to know your resources. For example, one
filesystem can differ very much from another. Otherwise, a
monitor interval is largely a matter of preference. Or your
particular needs for failover times. But you should also take
into account how much cpu is used by the monitoring operations. A
timeout should cover the worst case in which the resource may
still considered to be healthy.
> Joseph Lamoree
> Linux-HA mailing list
> Linux-HA at lists.linux-ha.org
> See also: http://linux-ha.org/ReportingProblems
More information about the Linux-HA