K8s cron job with bash scripts vs tool commands

[marrrcin]

Hi,
thanks for creating this tool!

I have a difficulty understanding the documentation related to backup/restore on Kubernetes. In the base example of doing backup/restore the tool is used by invoking... the tool itself:

shard_number=$(clickhouse-client -q "SELECT getMacro('shard')")
clickhouse-backup create_remote shard${shard_number}-backup
clickhouse-backup delete local shard${shard_number}-backup

while a few pages below, there is an example of a k8s cron job that actually does the INSERTs into the database which seems counter intuitive and "manual" so to say (and uses clickhouse-client)

clickhouse-client --echo -mn -q "INSERT INTO system.backup_actions(command) VALUES('create ${SERVER}-${BACKUP_NAMES[$SERVER]}')" --host="$SERVER" --port="$CLICKHOUSE_PORT" --user="$BACKUP_USER" $BACKUP_PASSWORD;

Is this just a leftover in readme or is there some particular reasoning why the tool cannot be run in cron job? Could you provide some context on that? It would help a lot. Thanks!

[Slach]

In kubernetes we use clickhouse-operator which deploy clickhouse-backup as container inside the Pod where already present clickhouse-server and execute clickhouse-backup server in REST API mode, which create engine=url() tables inside clickhouse-server

CronJob executes outside Pods with clickhouse-server + clickhouse-backup, and execute the same command via SQL queries

So I would like to recommend use clickhouse-backup cli interface for stanalone installations

And clickhouse-backup REST API for containerized installations

[Slach]

clickhouse-backup can't be executed inside CronJob cause it require call kubernetes API and could be disabled for security reasons

[marrrcin]

Thanks for a quick response! This make sense now alongside the 48:12 of this video https://youtu.be/megsNh9Q-dw?si=7q1EFsKfTawY0Ro4&t=2892 :)