Mrjob 步骤失败。怎么调试？

Posted 2023-04-18

【中文标题】Mrjob 步骤失败。怎么调试？

【英文标题】：Mrjob Step is failing. How do debug?

【发布时间】：2018-10-03 11:55:45

【问题描述】：

我正在尝试在 EMR 集群中运行示例 mrjob。我已经在 AWS 仪表板中手动创建了 EMR 集群并按如下方式启动了 mrjob

python keywords.py -r emr s3://commoncrawl/crawl-data/CC-MAIN-2018-34/wet.paths.gz --cluster-id j-22GFG1FUGS12L

作业因以下错误消息而失败

Using configs in /etc/mrjob.conf
Using s3://mrjob-07d6e1cbb9127021/tmp/ as our temp dir on S3
emr_api_params is deprecated and does nothing. Please use extra_cluster_params instead
Could not infer endpoint for bucket commoncrawl; assuming defaults
Copying local files to s3://mrjob-07d6e1cbb9127021/tmp/keywords.ec2-user.20181002.164319.430013/files/...
Adding our job to existing cluster j-22GFG1FUGS12L
Creating temp directory /tmp/phonenumers.ec2-user.20181002.164319.430013
Connect to resource manager at: http://localhost:40750/cluster
Waiting for Step 1 of 1 (s-2OZF2A4TZTS06) to complete...
  RUNNING for 0:00:18
  FAILED
Cluster j-22GFG1FUGS12L is WAITING: Cluster ready after last step failed.
Attempting to fetch counters from logs...
Waiting 10 minutes for logs to transfer to S3... (ctrl-c to skip)

如何查看失败消息？

【参考方案1】：

请参阅EMR docs 如何获取作业和任务日志。因为挖掘集群日志并非易事，我建议使用mrjob's local runner 详尽地测试 Python 代码。

【讨论】：

谢谢。我会检查。我无法使用 S3 在本地运行 mrjob（使用 boto python 库）。你有什么想法吗？