As a side-effect of T122638, the roots received the "shepherd traces" for job(s?) #501129:
```
12/21/2015 20:24:06 [600:7867]: wait3 returned -1
12/21/2015 20:24:06 [600:7867]: forward_signal_to_job(): mapping signal 20 TSTP
12/21/2015 20:24:06 [600:7867]: mapped signal TSTP to signal KILL
12/21/2015 20:24:06 [600:7867]: queued signal KILL
12/21/2015 20:24:06 [600:7867]: kill(-7869, TERM) -> overriddes kill(-7869, KILL)
12/21/2015 20:24:06 [600:7867]: now sending signal TERM to pid -7869
12/21/2015 20:24:09 [600:7867]: wait3 returned -1
12/21/2015 20:24:09 [600:7867]: forward_signal_to_job(): mapping signal 20 TSTP
12/21/2015 20:24:09 [600:7867]: mapped signal TSTP to signal KILL
12/21/2015 20:24:09 [600:7867]: queued signal KILL
12/21/2015 20:24:09 [600:7867]: kill(-7869, TERM) -> overriddes kill(-7869, KILL)
12/21/2015 20:24:09 [600:7867]: now sending signal TERM to pid -7869
12/21/2015 20:24:49 [600:7867]: wait3 returned -1
12/21/2015 20:24:49 [600:7867]: forward_signal_to_job(): mapping signal 20 TSTP
12/21/2015 20:24:49 [600:7867]: mapped signal TSTP to signal KILL
12/21/2015 20:24:49 [600:7867]: queued signal KILL
12/21/2015 20:24:49 [600:7867]: kill(-7869, TERM) -> overriddes kill(-7869, KILL)
12/21/2015 20:24:49 [600:7867]: now sending signal TERM to pid -7869
[…]
```
going on for 100k lines twice. I don't know what that means, but – once the dust of T122638 has settled – it would be nice to see if it can be avoided.