| Message ID | 20240206131655.32050-3-dwagner@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | fc transport cleanups | expand |
On 2/6/2024 5:16 AM, Daniel Wagner wrote: > When running the tests with FC as transport and the udev auto connect > enabled, discovery controllers are created and destroys while the tests > are running. This races with the cleanup code and also the > _find_nvme_dev() which iterates over all device entries and tries to > read the connect of transport and subsysnqn sysfs attributes. Since > these steps are not locked in anyway, the resources can go away in > between. > > Thus filter out 'cat' reporting non existing subsysnqn or transport > attributes. The tests will still fail if they can't fine the device etc. > But without filtering these errors out the tests fail randomly. > > Signed-off-by: Daniel Wagner <dwagner@suse.de> > --- > tests/nvme/rc | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/tests/nvme/rc b/tests/nvme/rc > index e0461f1cd53a..9cc83afe0668 100644 > --- a/tests/nvme/rc > +++ b/tests/nvme/rc > @@ -350,7 +350,7 @@ _cleanup_nvmet() { > > for dev in /sys/class/nvme/nvme*; do > dev="$(basename "$dev")" > - transport="$(cat "/sys/class/nvme/${dev}/transport")" > + transport="$(cat "/sys/class/nvme/${dev}/transport" 2>/dev/null)" do we have to do anything if there is in error ? > if [[ "$transport" == "${nvme_trtype}" ]]; then > echo "WARNING: Test did not clean up ${nvme_trtype} device: ${dev}" > _nvme_disconnect_ctrl "${dev}" > @@ -840,7 +840,7 @@ _find_nvme_dev() { > for dev in /sys/class/nvme/nvme*; do > [ -e "$dev" ] || continue > dev="$(basename "$dev")" > - subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn")" > + subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn" 2>/dev/null)" same here .. > if [[ "$subsysnqn" == "$subsys" ]]; then > echo "$dev" > fi -ck
On Tue, Feb 06, 2024 at 11:07:06PM +0000, Chaitanya Kulkarni wrote: > > for dev in /sys/class/nvme/nvme*; do > > dev="$(basename "$dev")" > > - transport="$(cat "/sys/class/nvme/${dev}/transport")" > > + transport="$(cat "/sys/class/nvme/${dev}/transport" 2>/dev/null)" > > do we have to do anything if there is in error ? In this case transport will be '' and not match with '${nvme_trtype}'. So we already handle this properly. > > if [[ "$transport" == "${nvme_trtype}" ]]; then > > echo "WARNING: Test did not clean up ${nvme_trtype} device: ${dev}" > > _nvme_disconnect_ctrl "${dev}" > > @@ -840,7 +840,7 @@ _find_nvme_dev() { > > for dev in /sys/class/nvme/nvme*; do > > [ -e "$dev" ] || continue > > dev="$(basename "$dev")" > > - subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn")" > > + subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn" 2>/dev/null)"
On Feb 06, 2024 / 14:16, Daniel Wagner wrote: > When running the tests with FC as transport and the udev auto connect > enabled, discovery controllers are created and destroys while the tests Nit: maybe you meant "destroyed" or "destroy" instead of "destroys". > are running. This races with the cleanup code and also the > _find_nvme_dev() which iterates over all device entries and tries to > read the connect of transport and subsysnqn sysfs attributes. Since > these steps are not locked in anyway, the resources can go away in > between. > > Thus filter out 'cat' reporting non existing subsysnqn or transport > attributes. The tests will still fail if they can't fine the device etc. Nit: s/fine/find/ > But without filtering these errors out the tests fail randomly. > > Signed-off-by: Daniel Wagner <dwagner@suse.de> > --- > tests/nvme/rc | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/tests/nvme/rc b/tests/nvme/rc > index e0461f1cd53a..9cc83afe0668 100644 > --- a/tests/nvme/rc > +++ b/tests/nvme/rc > @@ -350,7 +350,7 @@ _cleanup_nvmet() { > > for dev in /sys/class/nvme/nvme*; do > dev="$(basename "$dev")" > - transport="$(cat "/sys/class/nvme/${dev}/transport")" > + transport="$(cat "/sys/class/nvme/${dev}/transport" 2>/dev/null)" > if [[ "$transport" == "${nvme_trtype}" ]]; then > echo "WARNING: Test did not clean up ${nvme_trtype} device: ${dev}" > _nvme_disconnect_ctrl "${dev}" > @@ -840,7 +840,7 @@ _find_nvme_dev() { > for dev in /sys/class/nvme/nvme*; do > [ -e "$dev" ] || continue > dev="$(basename "$dev")" > - subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn")" > + subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn" 2>/dev/null)" > if [[ "$subsysnqn" == "$subsys" ]]; then > echo "$dev" > fi > -- > 2.43.0 >
diff --git a/tests/nvme/rc b/tests/nvme/rc index e0461f1cd53a..9cc83afe0668 100644 --- a/tests/nvme/rc +++ b/tests/nvme/rc @@ -350,7 +350,7 @@ _cleanup_nvmet() { for dev in /sys/class/nvme/nvme*; do dev="$(basename "$dev")" - transport="$(cat "/sys/class/nvme/${dev}/transport")" + transport="$(cat "/sys/class/nvme/${dev}/transport" 2>/dev/null)" if [[ "$transport" == "${nvme_trtype}" ]]; then echo "WARNING: Test did not clean up ${nvme_trtype} device: ${dev}" _nvme_disconnect_ctrl "${dev}" @@ -840,7 +840,7 @@ _find_nvme_dev() { for dev in /sys/class/nvme/nvme*; do [ -e "$dev" ] || continue dev="$(basename "$dev")" - subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn")" + subsysnqn="$(cat "/sys/class/nvme/${dev}/subsysnqn" 2>/dev/null)" if [[ "$subsysnqn" == "$subsys" ]]; then echo "$dev" fi
When running the tests with FC as transport and the udev auto connect enabled, discovery controllers are created and destroys while the tests are running. This races with the cleanup code and also the _find_nvme_dev() which iterates over all device entries and tries to read the connect of transport and subsysnqn sysfs attributes. Since these steps are not locked in anyway, the resources can go away in between. Thus filter out 'cat' reporting non existing subsysnqn or transport attributes. The tests will still fail if they can't fine the device etc. But without filtering these errors out the tests fail randomly. Signed-off-by: Daniel Wagner <dwagner@suse.de> --- tests/nvme/rc | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)