The first patch is addressing the problem, that the FC transport is way faster
in reconnecting and the test didn't observe all the states from live ->
resetting -> connecting -> live. Instead trying to see these transitions just
test for the final state which is live and the correct number of queues. This
makes this test also a little bit more robust. So this patch is necessary.
The next two patches are more in RFC state but I think it makes sense to post
them along side the rest.
The second and the third patch rely on the not yet released nvme-cli features
'volatile configuration' and 'execution context awareness'. These two feature
allow nvme-cli to figure out if a 'nvme connect' should actually be done or just
ignored. If the FC autoconnect udev/systemd rules are enabled on a host, this is
interfering with blktests. Note, this is also a way to get nvme-stas and
nvme-cli play nicely with each other.
In case anyone wants to run blktest with FC as transport needs either to disable
the autoconnect feature or use the unreleased features of nvme-cli.
Daniel Wagner (3):
nvme/048: Check for queue count check directly
nvme/rc: Avoid triggering host nvme-cli autoconnect
nvme/{041,042,043,044,045}: Use default hostnqn and hostid
tests/nvme/041 | 8 ++----
tests/nvme/042 | 8 ++----
tests/nvme/043 | 8 ++----
tests/nvme/044 | 8 ++----
tests/nvme/045 | 8 ++----
tests/nvme/048 | 35 ++++++++++++++++--------
tests/nvme/rc | 73 +++++++++++++++++++++++++++++++++++++++++++-------
7 files changed, 97 insertions(+), 51 deletions(-)
--
2.41.0
