DEV Community

Michael
Michael

Posted on • Originally published at gbase.cn

Monitoring Load and Export Task Progress in GBase 8a

#gbase #database #数据库 #monitoring

When you're moving data in and out of GBase 8a, keeping an eye on task progress is crucial. GBASE's MPP database offers great real‑time monitoring for load jobs, while export progress requires a slightly different approach. Here's how to check both.

Monitoring Load Tasks

For LOAD DATA jobs, the primary view is information_schema.load_status.

USE information_schema;
SELECT * FROM load_status;
Enter fullscreen mode Exit fullscreen mode

Key columns:

  • STATE – current status (RUNNING, FINISHED, FAILED, etc.).
  • PROGRESS – percentage complete; the most intuitive indicator.
  • AVG_SPEED – average throughput, useful for performance analysis.
  • ELAPSED_TIME – time spent so far.
  • LOADED_SIZE / TOTAL_SIZE – data volume perspective of progress.
  • LOADED_RECORDS, SKIPPED_RECORDS – row‑level progress and data quality.
  • DB_NAME, TB_NAME, DATA_SOURCE, SQL_CMD – task context.

Once a load finishes, load_status entries are removed. For historical results, query information_schema.LOAD_RESULT (per coordinator) or CLUSTER_LOAD_RESULT (cluster‑wide). To dig into failures, use SHOW GCLUSTER LOAD LOGS <task_id> for detailed error traces.

Monitoring Export Tasks

GBase 8a doesn't provide a dedicated progress view for SELECT ... INTO OUTFILE. Instead, you rely on:

  1. The process list: 
   SELECT * FROM information_schema.processlist
   WHERE COMMAND = 'EXECUTING' AND INFO LIKE '%INTO OUTFILE%';
Enter fullscreen mode Exit fullscreen mode

Check STATE (e.g., SENDING DATA) and TIME to see if the export is still running. However, there's no percentage progress available.

  1. File size at the destination: On the server where the output file is written, run ls -lh periodically. If the file is growing, the export is still underway.

Errors for export tasks are usually reported to the client that issued the command, or you can check express.log (under $GCLUSTER_HOME/log/gcluster/). For very large exports, consider splitting the output into batches and pre‑counting rows to estimate completion.

With these tools, you can stay on top of your gbase database ETL jobs — load monitoring is comprehensive, and export monitoring, while a bit more manual, is straightforward once you know where to look.

Top comments (0)