seekdb Management (obd seekdb)

seekdb is a lightweight database component managed by OBD. All obd seekdb commands only apply to deployments containing the seekdb component.

When to Use This Skill

• Installing or deploying seekdb instances
• Managing seekdb lifecycle (start, stop, restart, destroy)
• Setting up primary-standby replication
• Performing switchover, failover, or decouple operations
• Taking over existing seekdb instances not deployed by OBD

For OceanBase cluster management: Use cluster-management. For tenant operations: Use tenant-management.

Command Reference

| Command | Description | |---------|-------------| | obd seekdb install | Interactive install (single node, requires TTY) | | obd seekdb install --primary | Install as primary (enables RPC for standby sync) | | obd seekdb install --standby | Install as standby (selects from running primaries) | | obd seekdb deploy <name> -c <config> | Deploy with config file | | obd seekdb list | List seekdb deployments only | | obd seekdb start <name> | Start | | obd seekdb stop <name> | Stop | | obd seekdb restart <name> | Restart | | obd seekdb display <name> | Show info | | obd seekdb display <name> -g | Show topology graph | | obd seekdb destroy <name> | Destroy (see safety rules below) | | obd seekdb takeover <name> --home-path <path> | Takeover non-OBD instance | | obd seekdb switchover <name> | Planned primary-standby swap | | obd seekdb failover <name> | Emergency promote standby to primary | | obd seekdb decouple <name> | Decouple standby into independent primary |

Critical Safety Rules

Destroy with Standby

Destroying a primary that still has standby clusters requires --ignore-standby:

bash
obd seekdb destroy <name> --ignore-standby

Without this flag, OBD will refuse and warn about the risk.

Failover vs Switchover

| Operation | When to Use | Primary Must Be | |-----------|-------------|-----------------| | switchover | Planned maintenance | Online (running) | | failover | Primary is down/unreachable | Stopped or unreachable |

failover will error if the primary is still running. Use switchover for planned swaps.

Same-Host Restriction

Primary and standby must be on different IPs. If deployed on the same IP, OBD suppresses --role=STANDBY, causing the standby to start as a primary — log sync will not work.

Install Modes

See references/install-modes.md for detailed install mode behavior, TTY requirements, and non-interactive deployment limitations.

HA Operations

See references/ha-operations.md for detailed switchover, failover, and decouple procedures.

Usage Examples

Deploy and Start seekdb

bash
obd seekdb deploy my-seekdb -c seekdb-config.yaml
obd seekdb start my-seekdb

View Topology

bash
obd seekdb display my-seekdb -g

Takeover Existing Instance

bash
obd seekdb takeover my-seekdb --home-path /home/admin/seekdb

Related Skills

• cluster-management — OceanBase cluster lifecycle
• tenant-management — Tenant operations
• testing-and-benchmark — Performance testing