the plan doesn't show and the query won't execute

[GoogleCodeExporter]

I ran the following query from the command line, and it blocks (no output). I 
looked into the logs, but there isn't any useful message there. 
I stop-start the cluster (just to make sure I am not in a hanging mode) and 
rerun the query with all options checked but the "execute query" so I can see 
the plan. The plan won't show, and nothing in the logs. 
I try to see the plan on my local machine, and it shows up fine. Note that I 
have been testing this query in the past on 10 rows and it ran fine so I would 
think the problem is to run this query on the cluster (sensorium nodes 1..8). 
Investigating. 

//Query 12
//========

insert into dataset highest_value_page_per_user(
for $B in dataset page_views
where ( not(is-null($B.user)) and not(is-null($B.query_term)) )
group by $key_D := $B.user with $B
return {
 "group_key":  $key_D,
 "max": max(for $y in $B return $y.estimated_revenue)
}
);

for $l in dataset highest_value_page_per_user
return $l

insert into dataset total_timespent_per_term(
for $B in dataset page_views
where (is-null($B.user))
group by $key_alpha := $B.query_term with $B
return {
  "group_key":  $key_alpha,
  "sum": sum(for $y in $B return $y.timespent)
}
);

for $l in dataset total_timespent_per_term
return $l

insert into dataset queries_per_action(
for $B in dataset page_views
where ( not (is-null ($B.user)) and (is-null($B.query_term)))
group by $key_aleph := $B.action with $B
return {
  "group_key":  $key_aleph,
  "count": count($B)
}
);

for $l in dataset queries_per_action
return $l

Original issue reported on code.google.com by ker...@gmail.com on 18 Oct 2014 at 12:04

[GoogleCodeExporter]

More queries in the benchmark have a similar behaviour. 
I ran a count just to test the system & cluster and it ran fine.

Thanks,
Keren

Original comment by ker...@gmail.com on 18 Oct 2014 at 12:12

[GoogleCodeExporter]

Does it not show the plan on the Web-UI ? Maybe try to run the query on two or 
more logical nodes on your local machine (with "similar" data) and see if the 
plan gets generated. Or even better use physical nodes on the rainbow cluster 
to run your specific query to see if that helps, if YSK is done with his use of 
rainbow cluster.

Original comment by khfaraaz82 on 18 Oct 2014 at 12:38