Description of problem:
Starting in nvme cli rev 1.14 (~April 2020) the CLI was changed for the
connect-all command which is used to query discovery log pages.  Prior to 1.14,
connections were attempted for all discovery log page elements obtained from
the discovery controller.  As of 1.14, the cli was changed to only connect to
discovery log entries that had the same traddr (WWPN) as that of the discovery
controller.  This was done to "optimize" the number of connections attempted
due to another OEM's array which consistently returned log records for
subsystems on different FC ports on the subsystem, some of which may be on
different fabrics.   The implementation of this cli change has the traddr of
the discovery controller, which was built by the nvme-fc transport in the
kernel and passed to the nvme-cli request, compared against the content of the
discovery log record.  The kernel builds the traddr string using lower-case
hexadecimal characters. It appears that the Pure discovery log record returns
traddr using upper-case hexadecimal characters.  Note: there is no statement in
the NVME-FC standard that it must be upper or lower so there is no standards
violation. When nvme-cli compares, it is not case-insensitive, thus the
kernel-created traddr does not match the PURE log traddrs and no connections
are attempted.


The fix to this issue has been submitted upstream and incorporated into nvme-cli, but the cli release has not been created. 

nvme-cli: nvmf-connect@.service: Remove matching from default syntax
https://github.com/linux-nvme/nvme-cli/commit/58c23ceb12d16756b2222a55d1d9dc5f34bb4fa4

https://github.com/linux-nvme/nvme-cli/commit/1264c6323937c4a0342174fdd9be5a66ab1eaf24
nvme-cli: Make connect-all matching be case insensitive

Both of these patches correct the issue. However, it is recommended that the first patch, removing "--matching" from the default connect service file be pulled in. This restores behavior of prior 8.x releases for the scripts.