Query doesn't seem to be attempted

[#1]

Hi,

We've switched to the latest AWS EMR version (4.1) which includes Hue 3.7.1.  Some queries work but certain ones don't appear to even start. For example:

select rcd_seq_num
from comverse.zzz_ocs
where event_start_yearmonth = 201401 --partition
and market_offer_group_id = 113 --partition
limit 100;

When you press Execute it flashes red quickly and then goes back to Execute. After after 2-3 minutes this message appears at the bottom:

Fetching results ran into the following error(s):

timed out

Looking in the hive logs it does appear that it is searching through S3. Is there a setting in Hue to increase any timeouts?

[#2]

If it helps, this works fine on emr 3.9.

Interested to know what causes the Execute button to stay red (as on 3.9) or flick to red and then back to blue (as on 4.1)

[#3]

If you submit with beeline on the CLI, do you have the same problem?
Anything in the HiveServer2 logs?

(this is probably unrelated to Hue but more a Hive setup problem)

[#4 · Accepted solution]

Hi Romain,
Spent an hour with an AWS engineer and cracked it. It did return via Beeline on the CLI after about 5 mins. We had to modify the Beeline timeout in hue.ini to a much higher value than 120.

I suspect it spends most of that initial time scanning S3.

[#5]

Ha, hope Hive will speed-up in the next version 😉