Troubleshooting OceanBase NTP Clock Synchronization Issues with ChatDBA

Problem Description

The OBServer instance of OceanBase fails to start because its NTP clock synchronization is abnormal, leading to a server startup error.

Interactive Experiment with ChatDBA

Round 1

ChatDBA receives the problem description and generates an initial reference result, prompting the collection of journalctl logs for further analysis.

Round 2

After providing the journalctl output, ChatDBA points out errors related to ntpdate configuration and DNS settings, suggesting inspection of the ntpdate-wrapper script.

Round 3

Based on the supplied ntpdate configuration, ChatDBA narrows the cause to the ntpdate-wrapper script, recommending a review of its contents.

Round 4

Investigation of the script reveals that it imports ntp.conf and references a step‑tickers file, which contains the server address 0.kylin.pool.ntp.org matching the log error, confirming the root cause.

Final Resolution

ChatDBA confirms that the misconfiguration in the step‑tickers file caused the NTP synchronization failure and provides a corrective solution, which resolves the issue.

Experiment Summary – Common NTP Sync Failure Causes

NTP Service Configuration Issues: Incorrect or unreachable NTP server addresses in ntp.conf can prevent the service from starting.

DNS Resolution Problems: Failure to resolve hostnames (e.g., "Can't find host 0.kylin.pool.ntp.org") leads to sync errors.

Configuration File Errors: Errors in ntpdate-wrapper or step‑tickers may cause the service to use wrong servers.

Network Problems: Unstable or misconfigured network connections hinder communication with NTP servers.

Permission Issues: Insufficient privileges for the NTP service can block startup.

Clock Source Problems: Virtual machines without a stable clock source may experience drift.

Incorrect Service Mode: NTP must run in slew mode ( -x ) to handle large offsets gracefully.

NTP Service Configuration Details

The ntp.conf file lists multiple time servers for fallback, while the optional step‑tickers file specifies servers for large time offsets used by ntpdate . In this case, an outdated entry in step‑tickers caused the failure, so commenting out or correcting that entry resolves the issue.

Additional Community Announcements

The article also includes a preview of upcoming community events, such as the PyCon China 2024 schedule, and links to previous "One Question One Experiment" issues (51‑58) for further reading.