> Samuel, > > I was trying out some quick things to root-cause the problem > and I noticed that mtu_reboot test sometimes gets stuck. I see that > the test does this > > tester ----> Test machine > > tester remotely (via ssh/rsh) executes dladm call on test_machine > and then reboots test_machine. After that, the tester does a > wait_for_reboot. > > When I tried this today, the wait_for_reboot does not return, > even though the test_machine is up (and I can ping it, and > do "rsh -n test_machine echo good" on it). > > Is there some bug in wait_for_reboot? Is it testing for the correct > return status from ping/rsh? > > When I was testing this , it failed to set the mtu to 1000 . And I suspect there's something uncertain in the behavior of this setting mtu. However I'll look into this. > BTW, we will also need to add tests for flowctrl, similar to > mtu (though reboot will not be needed to test this). > I am considering using a more generic framework of testing this so that a list of values could be easily tested with the framework . I've completed it , not yet tested.
Samuel
