Hi, Mingyu Chen!
Thank you for your advice! I did actually help, now I have about 95%
of HDFS throughput using Hive.
Kind regards,
Nikita
пт, 10 мар. 2023 г. в 18:37, Mingyu Chen :
> Hi Nikita,
> First of all, you don't have to set all those configs and session
> variables.
> Setting them does
Hi Nikita,
First of all, you don't have to set all those configs and session variables.
Setting them doesn't do anything for this problem.
Let me make it clear, you problem is that:
1. Use `insert into table from hdfs()` can utilize most of your cluster's
resource, and the speed is acceptable.
2
Hi, Yongqiang YANG, thank you for your response!
I appended fe.conf with lines:
async_load_task_pool_size = 320
async_loading_load_task_pool_size = 320
async_pending_load_task_pool_size = 320
320 is the total amount of cores in my configuration.
I had restarted all the machines in the cluster (i
You can adjust async_pending_load_task_pool_size = 250
and async_loading_load_task_pool_size=250 in fe.conf and have a try.
On Fri, Mar 10, 2023 at 8:49 PM Nikita Sakharin
wrote:
> Hi, Apache Doris Community!
>
> I'm a beginner user of Doris and I'm experiencing performance issues.
>
> I have *4
Hi, Apache Doris Community!
I'm a beginner user of Doris and I'm experiencing performance issues.
I have *4 hosts for BE: Intel Xeon 80 cores, 768GB RAM each.* Frontend and
broker were deployed on a different machine with the same configuration.
All backends have "*mix*" as *Node Role*.
Doris ver
