Re: Hive queries rejected under heavy load

[Stephen Sprague]

gotta start by looking at the logs and run the local client to eliminate
HS2.   perhaps running hive as such:

$ hive -hiveconf hive.root.logger=DEBUG,console

do you see any smoking gun?

On Wed, Sep 28, 2016 at 7:34 AM, Jose Rozanec  wrote:

> Hi,
>
> We have a Hive cluster (Hive 2.1.0+Tez 0.8.4) which works well for most
> queries. Though for some heavy ones we observe that sometimes are able to
> execute and sometimes get rejected. We are not sure why we get a rejection
> instead of getting them enqueued and wait for execution until resources in
> cluster are available again. We notice that the connection waits for a
> minute, and if fails to assign resources, will drop the query.
> Looking at configuration parameters, is not clear to us if this can be
> changed. Did anyone had a similar experience and can provide us some
> guidance?
>
> Thank you in advance,
>
> Joze.
>
>
>

Hive queries rejected under heavy load

[Jose Rozanec]

Hi,

We have a Hive cluster (Hive 2.1.0+Tez 0.8.4) which works well for most
queries. Though for some heavy ones we observe that sometimes are able to
execute and sometimes get rejected. We are not sure why we get a rejection
instead of getting them enqueued and wait for execution until resources in
cluster are available again. We notice that the connection waits for a
minute, and if fails to assign resources, will drop the query.
Looking at configuration parameters, is not clear to us if this can be
changed. Did anyone had a similar experience and can provide us some
guidance?

Thank you in advance,

Joze.