Your questions answered!

Question of the Week

Parquet can only be used with s3() table function. Tiered disks can only work with MergeTree. One of potential ideas for ClickHouse is to have MergeTree over Parquet, i.e. use Parquet as an underlying data format instead of internal ClickHouse one. This may come in the future.

Past Questions

Yes, but both storage policies should be compatible (have all disks/volumes).

The easiest way is to reduce the memory you are using by:

  • Simplifying the query to do as much as possible using scans.
  • Reduce the number of rows selected by pushing filters as far down as possible.
  • Reducing the number of columns selected in the scan
  • Reducing the amount of memory used by joins (switch to Dictionary if appropriate)
  • Reducing the memory in aggregates by using cheaper, less expensive forms.

You may define specific grants in users.xml.

<clickhouse>
   <users>
      <default>
         <access_management>1</access_management>
         <grants>
            <query>GRANT SHOW SETTINGS PROFILES ON *.*</query>
         </grants>
     </default>
   </users>
 </clickhouse>

There are no explicit limits. In a cluster, it’s only stored in ZooKeeper if you enable it explicitly. Normally each RBAC account is stored locally on disk. You can use ‘create user on cluster’ in the latter case.

Yes!

Have other burning questions about ClickHouse?

Submit them!

Want to chat with an expert? Book a free consultation