Using ODBC drivers about clickhouse-operator HOT 4 CLOSED

Some clarification is needed. Would you like to add driver on every cluster's node? Why not to add driver on the client's side?

from clickhouse-operator.

@sunsingerus we need to have reciepe for that.

Most probably can be done as
A) install unixODBC package
B) install needed odbc driver via package/archive (we can pick one as a sample) - it can be postgres, Oracle, sql server etc.
C) register odbc DSN using a template provided by user and using odbcint tool

Check that
https://github.com/ClickHouse/ClickHouse/pull/9348/files#diff-62c528ac2c8fb7f5282ea76540fe63b6
(obviously instead of clickhouse odbc driver we need some other in sample, because clickhouse usually need to connect to some external sources via odbc, not to other clickhouse)
Similar - for jdbc.

from clickhouse-operator.

+1 for this..

Can you post some information on how to add the shared library for the ODBC driver and the odbc.ini file in the right spot so the ODBC table engine can pick it up?

Thanks!

from clickhouse-operator.

@a-dot you need custom image which contains required ODBC driver
look Docker file examples on https://gist.github.com/9f9449a722091a13a9069b79f8dc7da7
build and publish custom image with installed ODBC drivers

after that
look kubectl explain chi.spec.configuration.templates.podTemlates and kubectl explain chi.spec.configuration.clusters.templates, kubectl explain pod.spec.template.containers and try to figure out how to create podTemplate with container with your custom image and link it to cluster pods inside kind: ClickHouseInstallation resource

from clickhouse-operator.