HyperFabric Release Notes (B.11.11.04.02)

HyperFabric Release Notes
Patches and Fixes in This Version
Chapter 114
CR JAGaf21261 Node is not marked down now, if the VC establishment logic fails
prematurely when VC info could not be sent out due to non blocking
TCP send failures (EAGAIN).
CR JAGaf21262 Send was being marked successful for failed send if VC message id says
send is done. It did not distinguish this earlier. Oracle could crash in
this case only during failure events.
CR JAGaf21264 CLIC commands were not reflecting the What string. Included the
linking of revision.o for all commands.
CR JAGaf21278 SLOS messages are populated in the SKGXP layer to take care
providing the necessary information in the error path (proactive
change - no reported adverse effects).
CR JAGaf21269 skcsisndwt() is done using unified queue for both sends/receives
(proactive change - no reported adverse effects).
CR JAGaf21283 cancel outstanding sends --or-- put retries will attempt multiple times
(10) with an exponential back off, to take care of possible connectivity
losses and retransmission of MSGIDU. Earlier we could prematurely
gave up retrying on a failure event which could result in Oracle crash.
CR JAGaf21289 If clic_send_wait returns CLIC_INTR then SKGXP returns now
SKGXP_POSTED. Earlier it was returning SKGXP_FAIL which was
resulting in Oracle crash.
CR JAGaf21293 Retransmit queue is traversed in correct direction while retransmission
after a failover event. It was earlier resulting update sequence/out of
order retransmission causing Oracle to crash.
Table 1-3 Fixes Included in this Release (Continued)
CR Number Description