HP Netserver MSCS SPOFless Cluster Solution

CasualtyProcedures
Conclusion
!
!
!
!
!
FailedFCArrayController
FailedSCSICable/Controller
IIntheeventofanFCcontrollerfailure,thefollowingstepsshouldbetaken:
Obtainanewcontrollerandupgradethefirmwaretothesameversionoffirmwareasthe
workingFCcontroller.
Performtheupgradeona
separateserverwithanRS/12FC.DonotuseaclusteredRS/12FC.
RemovethefailedcontrollerfromtheRS/12FC.
Attachthecablestothenewcontroller.
Insertthenewcontroller.Thelightonthebackofthenewcontrollershouldblinkyellow
for1-5minutes,thenthelightshouldturngreen,andtheRS/12FCshouldreturnto
normaloperation.
Verifycontroller'sparametersettingsusingtheFCArrayAssistant.
Ifthefirmwarebetweenthetwocontrollersdoesnotmatch,the
storagefunctionalitywillbedisruptedandtheclusterwillfail.
IntheeventofafailedSCSIcableorafailedSCSIcontroller,thefollowingstepsshouldbe
taken:
ObtainanewSCSIcableorcontroller.
DisengagetheFCcontrollerthatisblinkingredfromtheRS/12FC.
Replacethefailedcableorinsertthenewcontroller.Thelightonthebackofthenew
controllershouldblinkyellowfor1-5minutes,thenthelightshouldturngreenandRS/12
FCshouldreturntonormaloperationonit'sown.
Verifycontroller'sparameterssettingusingFCArrayAssistant.
TheSPOFlesssolutionsdescribedinthisdocumenthavepassedallphasesoftheMicrosoft
Hardware CompatibilityTestsinconfigurationsthatmeetallMicrosoftrequirements.Inthefuture,
theHBAdriversareexpectedtosupportmultipleFCloopssothatmorethanoneHBAcanbe
installedintheclusterservers,providingredundantpathstothestorage.Currentcluster
solutionswillthenbecomeupgradable,supportingasecondHBAineachserver.Redundant
HBAswillcreateahigherlevelofavailabilitybyrepresentingatrulyredundantcluster.The
diagramonthefollowingpageshowstheSPOFlessconfigurationmeetingallredundant
requirements(redundantnetworkadaptersfor publicandprivatenetworksarenotshown).
!
!
!
!
HP
DocumentationNetServer
whitepaper
8