As I wrote, the script file assumes that if myts isn't running, then the pipe node needs to be created.
Since myts does not seem to set any trap handling (I.E: assumes it will always do a normal exit, not a trap exit) then the above assumption is incorrect for the case of anything other than a normal exit path.
Translation: The segfault prevents it from running, the less than robust script logic makes mknode fail on a pre-existing pipe. (Three problems: the script logic, the (lack) of trap logic and the original cause of the segfault.)
The strace output should help in finding the third problem. Only the designer can find the first two.