Resilient Team operation not detecting dead Team member

Description

Team operations (e.g. barrier) sometimes succeed even when a team member has died before the start of the operation.

This behavior occurs intermittently in x10.tests/tests/Resiliency/TestBarrier.x10 .

[noformat]
jjmiltho@sussman:~/x10/x10.tests/tests/Resiliency$ for i in `seq 1 30`; do X10_RESILIENT_MODE=1 X10_NPLACES=25 ./a.out 23; done
ERROR root setting the second child Place(13):team0 to PHASE_DONE
Place 23 exited unexpectedly with exit code: 1
Launcher 23: cleanup complete, exit code=1. Goodbye!
++++++ Test succeeded.
Place(0) passed second barrier!
Place(1) passed second barrier!
Place(7) passed second barrier!
Place(8) passed second barrier!
Place(9) passed second barrier!
Place(2) passed second barrier!
Place(5) passed second barrier!
Place(10) passed second barrier!
Place(11) passed second barrier!
Place(3) passed second barrier!
Place(12) passed second barrier!
Place(4) passed second barrier!
Place(6) passed second barrier!
Place 23 exited unexpectedly with exit code: 1
Launcher 23: cleanup complete, exit code=1. Goodbye!
++++++ Test failed.
...
[noformat]

Environment

Resilient X10

Status

Assignee

Unassigned

Reporter

Josh Milthorpe

Labels

None

External issue ID

None

Affects versions

X10 2.6.0

Priority

Medium
Configure