In mmp.sh, there is no check_failover_pair in the beginning of test 8-10, which might create some fake results when there is no failover-setup there. Though I am not so sure about this. Can some one please check that?
For test 8-10, they are cases for verifying running e2fsck on Lustre server target which has MMP feature enabled:
test 8: mount during e2fsck (should fail)
test 9: mount after aborted e2fsck (should fail)
test 10: e2fsck -n with mounted filesystem (should pass)
without failover partner, the above three test cases should also work. So, check_failover_pair was not used in them.
|