How can we realistically manage the extra processes required for ColumnStore to function?
In a regular scenario a single node on a VM or bare metal installation uses systemd to run ColumnStore.
The systemd unit file, mariadb-columnstore.service, calls /usr/bin/mariadb-columnstore-start.sh.
When setting up a cluster, cmapi takes over and handles the processes instead.
To be able to run CS in docker, we also need to handle these processes.
Options:
Simply run the process in the backround. This is what we currently do, but more of the scripts like columnstore-post-install need to be replicated.
supervisor, same as above.
Use cmapi even for one node. The same issue persists, more setup needs to be rewritten
How can we realistically manage the extra processes required for ColumnStore to function?
In a regular scenario a single node on a VM or bare metal installation uses systemd to run ColumnStore. The systemd unit file, mariadb-columnstore.service, calls /usr/bin/mariadb-columnstore-start.sh.
When setting up a cluster, cmapi takes over and handles the processes instead.
To be able to run CS in docker, we also need to handle these processes.
Options:
columnstore-post-installneed to be replicated.