Fujitsu V3.1 Manual De Usuario
To allow the switchover in this case, VIOM provides a “forced unassign”
option, allowing you to unassign a profile even if the corresponding MMB (con-
taining this profile) is not available.
option, allowing you to unassign a profile even if the corresponding MMB (con-
taining this profile) is not available.
But applying this function means that there will be two chassis with at least
partially identical profiles, potentially resulting in, for example, duplicate vir-
tual addresses as soon as the failed site goes online again.
partially identical profiles, potentially resulting in, for example, duplicate vir-
tual addresses as soon as the failed site goes online again.
This chapter describes the settings and procedures for avoiding duplicate
addresses and how to handle a site failover with VIOM.
addresses and how to handle a site failover with VIOM.
As already mentioned, there are two major issues: unassign the profiles at a
failed / no longer reachable site, and handle the power-on procedure of the
failed site to avoid duplicate MAC and WWN addresses.
failed / no longer reachable site, and handle the power-on procedure of the
failed site to avoid duplicate MAC and WWN addresses.
The first issue is solved by the “forced unassign” function, which is available
as a menu item in the VIOM Web GUI as a CLI command.
as a menu item in the VIOM Web GUI as a CLI command.
But beware: This function should only be used if there is no other
way to get the profile unassigned! A forced unassign will only
remove the “profile-to-server blade assignment” in the VIOM data-
base but not in the corresponding hardware (MMB, IBP, server
blade). If you later assign this profile to a different server blade you
may end up with two different blades containing identical virtual
addresses.
way to get the profile unassigned! A forced unassign will only
remove the “profile-to-server blade assignment” in the VIOM data-
base but not in the corresponding hardware (MMB, IBP, server
blade). If you later assign this profile to a different server blade you
may end up with two different blades containing identical virtual
addresses.
Therefore you should only assign and boot the profile on the other site if you
are sure that the failed site is not longer active on the networks. The easiest
way to ensure this is to completely power-off the failed site before restarting
the profiles on the other site.
are sure that the failed site is not longer active on the networks. The easiest
way to ensure this is to completely power-off the failed site before restarting
the profiles on the other site.
To avoid having two blades powered-on with the same server profile, you
must prepare the power-on behavior of the chassis. During site preparation
the server blades’ BehaviorAfterACFail flag need to be set to Always
power off. You can do this in either of the following ways:
must prepare the power-on behavior of the chassis. During site preparation
the server blades’ BehaviorAfterACFail flag need to be set to Always
power off. You can do this in either of the following ways:
Via MMB: (preferred because more convenient)
The MMB “Power Restore Policy” function on the Power Management tab
controls the “Behavior After AC Fail”. Set it as indicated to Always power
off for every server blade in the chassis.
controls the “Behavior After AC Fail”. Set it as indicated to Always power
off for every server blade in the chassis.
ServerView Virtual-IO Manager
335
13.3 Disaster Recovery