so if it is a timing race, does it really take longer than 8 seconds for that target to no longer be active? Or is there something wrong where it thinks that target is the one being used between volume1 and server2 and we haven't yet deleted server2?
I'm not sure that a retry in cinder will fix the problem since the attachment delete for server1/volume1 fails here:
http:// logs.openstack. org/81/ 606981/ 4/check/ tempest- slow/fafde23/ controller/ logs/screen- c-vol.txt. gz#_Dec_ 08_01_45_ 41_380161
and the volume delete fails for the same reason 8 seconds later here:
http:// logs.openstack. org/81/ 606981/ 4/check/ tempest- slow/fafde23/ controller/ logs/screen- c-vol.txt. gz#_Dec_ 08_01_45_ 49_587081
so if it is a timing race, does it really take longer than 8 seconds for that target to no longer be active? Or is there something wrong where it thinks that target is the one being used between volume1 and server2 and we haven't yet deleted server2?