RadiSys ATCA-4616 Dokumentacja Pobierz pdf (Strona 107)

Troubleshooting

107

EachIPMCthatisnottheactiveShMCmonitorsits“ownside”ofthebus.Inotherwords,

eachIPMCismonitoringitsownI

CinterfacesthatareconnectedtoIPMB‐AorB.Periodically

itchecksforerrorsontheI

CinterfacesusingregistersintheIPMC.Ifanerrorisdetected,the

IPMCisolatesitselffromtheIPMB,resetsitself,andtriestorejointhebus.IftheIPMCisnot

successful,itgeneratestheIPMB‐0sensoreventviatheotherbusandkeepsitselfisolated

fromtheerror‐pronebus.TheIPMCperiodically(every250msecs)triestoresettheI

C

interfaceandrejointheerror‐pronebusuntiltheerrorcorrects.

TheactiveShMChastheresponsibilityofmonitoringnotonlyitsownsideoftheIPMBbut

alsotheexternalIPMB.Itcanthusdetectifsomeothererror‐proneIPMCinthechassisis

holdingoneoftheIPMBsdown,andisolateitselffromthatbus.Alltransactionsarethen

doneontheotherbus.PeriodicallytheShMCwilljointheerror‐pronebustotestiftheerror

isstillthere.

IPMB and IPMC failure problems

AllIPMCsarbitratetogetaccesstothebus,butmessagescanbemissedifthereisalotof

trafficonthebus(e.g.runningastresstest),andallmessagingbuffe rsontheIPMCarefull.

EachIPMCretriesrequeststhreetimesandtheShMSretrieseachrequest5times.Also,hot‐

swapeventsareguaranteedtobereceivedbecausetheygetretrieduntiltheShMS

acknowledgesthem.Thisensuresthatnohot‐swapeventislost.TheIPMC'seachhavea50

messagebuffer,andtheShMChasroomforanadditional10messages.Thechanceof

overrunningthemessagebuffersislow,butitispossibleifeverybladeinthechassis

simultaneouslygenerateseventsfrommultiplesensors.

MessagecorruptioncanhappenontheIPMBunderlimitedcircumstances,suchaswhenthe

IPMCisreceivingandformattingthemessageduetoalocalIPMCerror.Ifthemessageis

corrupted,theIPMCchecksboththeheaderandthemessagechecksum.Ifeithertheheader

orchecksumisincorrect,themessageisdropped.ThesendingIPMCthenretriesthe

message.Thereislittlechanceofthesamemessagegettingcorruptedonthebusthreetimes

insuccession.

If IPMB-0 Isolation events are seen frequently in Event Log

Thefollowingeventsmighthaveasimilarcause:

•AFRUsendsfrequentIPMB‐0Isolationevents

•TheactiveSCMreportsfrequentIPMB‐0Isolationevents

•TheShMSisunabletocommunicatewithshelfIPMCsbecauseoffrequentIPMB‐0

Isolationissuesintheshelf

1 2 ... 102 103 104 105 106 107 108 109 110 111 112 ... 145 146

Brak uwag

RadiSys ATCA-4616 Dokumentacja Strona 107