databend-query cluster run queries only on one node?

[Lif0820]

I have a databend-query cluster with 3 nodes, which can be found in table system.clusters:
name host port
HnMW... 10.66.110.37 9091
opIBW... 10.66.110.39 9091
qJHdo... 10.66.110.38 9091

I send a slow query sql (named sql_A) to any one node (e.g. 10.66.110.38:3307) above by mysql client,
and see cpu usage on linux shell by top command, only 10.66.110.37 cpu become busy, and the sql result cost about 10 seconds;

next, i send the sql_A query twice at same time, it stiil seems to only use 10.66.110.37 cpu, and the sql result cost almost 20 seconds;

so, my question is:

1. the databend-query cluster has 3 nodes, why it seems like only the first node in table system.clusters will be used?
2. for one sql query which is sended to databend-query cluster, it will use only any one query-node? or may use all nodes to complete the query like Greenplum?

[wubx]

databend-query process query is cluster mode.
There are some algorithms:

1. if query user segment > cluster_size , every databend-query process segment/cluster_size
   table segment can use: select * from fuse_snapshot('db','table_name') limit 1;

2. if segment< cluster_szie , every databend-query processs blocks/cluster_size
   select * from fuse_snapshot('db','table_name') limit 1;

if the databend-query onlye one node process we need to make sure is there one segment and include one block.