TDengine taosd Configuration Guide for Stable Time-Series Workloads

What the taosd process does

The taosd process is the core TDengine server process. It handles writes, query execution, log processing, monitoring, and other database operations for time-series workloads.

By default, TDengine reads configuration from /etc/taos/taos.cfg. You can point taosd to another configuration directory from the command line when deployment layouts require it.

taosd command-line parameters

The taosd service supports several key command-line flags:

• -c: config directory
• -C: print config and exit
• -e: env file path
• -o: log output mode (stdout or file)
• -V: print version
• -r: repair mode

Version and config viewing: taosd -V shows the version; taosd -C prints all active configuration parameters.

Custom config directory: taosd -c /data/tdengine/config and combining flags: taosd -c /data/tdengine/config -e /data/tdengine/env. The config directory requires taos.cfg plus any needed certificate files.

taosd repair mode

Started via taosd -r, repair mode handles three file types:

• meta files: database metadata (table structures, supertables)
• tsdb files: time-series data storage
• WAL files: write-ahead log storage for reliability and disaster recovery

Typical use cases include recovery after abnormal shutdown (WAL files may contain unflushed data), disk sector corruption repair (tsdb files), and metadata consistency checking.

Log check command: tail -f /var/log/taos/taoslog0.log

Core configuration parameters

Connection-related:

• firstEp: required, cluster entry node
• secondEp: failover address
• fqdn: fully qualified domain name for client connections
• serverPort: default 6030

For production clusters, use stable endpoint planning. Static IPs can reduce name-resolution risk, while well-managed DNS may still be appropriate in controlled environments.

Monitoring-related:

• monitor: enable (1) or disable
• monitorFqdn: target address for push
• monitorPort: default 6043
• monitorInterval: collection interval in ms, recommended 10-30 seconds

Query-related:

• tagFilterCache: 256 MB default, caches intermediate tag filter results
• queryRspPolicy: file (disk only), memory (memory only), or both (memory-first, fallback to disk)

Production configuration recommendations

High-concurrency writes:

• bufferPoolSize: 256
• walFsyncDuration: 100
• maxConnections: 5000

Large-scale queries:

• numOfQueryThreads: 16
• queryRspPolicy: memory
• tagFilterCache: 512

Troubleshooting

Service startup failure: Check port usage with netstat -tlnp | grep 6030, view error logs at /var/log/taos/taoslog0.log, and validate config with taosd -c /etc/taos -C.

Connection refused: Verify fqdn, check firewall rules, and confirm firstEp/secondEp addresses are reachable.

Summary

taosd configuration has a direct impact on write throughput, query behavior, monitoring, and recovery. Start with conservative production settings, then tune parameters against real workload patterns and operational metrics.