Job stays at RUNNING state after END of a churn trace

[heversonbr]

It seems that a job running with a churn trace does change its state to ENDED when it finishes. Although the end of trace is logged at the controller the job keeps its RUNNING state when checked with "splay-list-jobs.lua" . See outputs below:

I, [2016-03-10 22:02:54 #20880] INFO -- : Job 125: RUNNING
I, [2016-03-10 22:03:19 #20880] INFO -- : TRACE 125: time 21
I, [2016-03-10 22:03:19 #20880] INFO -- : TRACE 125: stop 4
I, [2016-03-10 22:03:22 #20880] INFO -- : TRACE 125: time 23
I, [2016-03-10 22:03:22 #20880] INFO -- : TRACE 125: stop 2
I, [2016-03-10 22:03:44 #20880] INFO -- : TRACE 125: time 43
I, [2016-03-10 22:03:44 #20880] INFO -- : TRACE 125: stop 3
I, [2016-03-10 22:03:53 #20880] INFO -- : TRACE 125: time 50
I, [2016-03-10 22:03:53 #20880] INFO -- : TRACE 125: start 2
I, [2016-03-10 22:04:05 #20880] INFO -- : TRACE 125: time 60
I, [2016-03-10 22:04:05 #20880] INFO -- : TRACE 125: start 6
I, [2016-03-10 22:04:07 #20880] INFO -- : TRACE 125: time 62
I, [2016-03-10 22:04:08 #20880] INFO -- : TRACE 125: start 4
I, [2016-03-10 22:07:29 #20880] INFO -- : TRACE 125: time 241
I, [2016-03-10 22:07:29 #20880] INFO -- : TRACE 125: stop 1
I, [2016-03-10 22:07:29 #20880] INFO -- : TRACE 125: stop 2
I, [2016-03-10 22:07:29 #20880] INFO -- : TRACE 125: stop 4
I, [2016-03-10 22:07:29 #20880] INFO -- : TRACE 125: stop 5
I, [2016-03-10 22:07:29 #20880] INFO -- : TRACE 125: stop 6
I, [2016-03-10 22:07:29 #20880] INFO -- : END OF TRACE

splayd@ubuntu:⟫ date
Thu Mar 10 22:10:38 UTC 2016

splayd@ubuntu:⟫ ./splay-list-jobs.lua

Job List =
job_id=121, user_id=1, status=KILLED
job_id=122, user_id=1, status=KILLED
job_id=123, user_id=1, status=KILLED
job_id=124, user_id=1, status=KILLED
job_id=125, user_id=1, status=RUNNING

Totals =
REGISTER_TIMEOUT = 0
QUEUED = 0
NO_RESSOURCES = 4
ENDED = 3
LOCAL = 0
KILLED = 117
REGISTERING = 0
RUNNING = 1

[vschiavoni]

is this happening with any job with churn, or only those with --trace_alt ?

[heversonbr]

I've been doing tests with --trace_alt only this moment, Ill check for both cases when I'll handle more closely this issue.