User guide

DataDirect Networks SFA™ OS 2.0.0 Release Notes Revision A1 | 10
A controller crash occurred when the shutdown subsystem abrupt command
was issued on controller 0 and then again on controller 1. This was due to an issue
where the second controller forwarded the second command to the master controller
while it was trying to shutdown.
• SFA12K controller crash with JEX_Storage_EnqueueRequest
This issue was due to a problem in the logging agent inside the SFA OS.
• Latency on SSD virtual disks without corresponding physcial disk latency
High levels of latency were observed under certain workloads in the use of SSD
pools/VDs without an associated PD latency. This was due to small reads getting
blocked waiting for buffers due to the large read workload ahead of it.
• LOG_ES_SES_COMMAND_QUERY_FAILURE repeating every 30 seconds
This issue was reported on an older SFA-10K with AAMUX dongles. This was due to
path changes on the SES device not being registered correctly.
• After DEM upgrade the VDs continue to all stay on one controller
This issue occurred when the secondary controller is in MIR state NO BACKEND
DRIVES and the primary controller is activating the JIS.
• Show subsystem summary all shows bad block numbers with negative values
The show subsystem summary all would show bad block numbers with negative
values while the 'show vd X all' command, the bad block count shows up as zero
instead of the large negative number. This was a reporting issue caused by a problem
in the SFA OS.
• Show internal_disk incorrectly reports disks as not present
A reporting issue with the show internal disk command was reported where disks
were reported as not present when they were present. This was due to an issue in
SFA OS where stale data was used by the reporting function.
• Initiator logout and subsequent attempt to abort outstanding I/O caused controllers
running SFA OS 1.5.3.0 to crash. This was due to a bug in the SCSI persistent
reservation code that accessed memory beyond its array size. (DE9111)
• Inside a VM the driver for Mellanox ConnectX3 and ConnectX2 HCAs failed to load
with error message "SW2HW_MPT failed."
• Previously, binding two or more Mellanox HCAs to a VM often showed one less.
• Previously, when you set a stack's CD_DVD_DEVICE attribute, you would have to
unset it so that the VM doesn't attempt to re-install itself a second time. This is no
longer necessary. The SFA controller now detects when the operating system ejects
the DVD and then disassociates it automatically from the stack. ISO installers,
including those for GRIDScaler and EXAScaler, eject the DVD at the end of
installation. [DE7427]