分布式训练，ZK连接失败
- Dominant language
- C++
- Stars
- 2.9k
- Forks
- 553
- PR merge metrics
- No merged PRs in 30d
Description
离线内网环境。采用制作docker镜像的方式部署欧拉，共有3台虚拟机（10.200.33.4 和10.200.33.5和10.200.33.6），1台作为ps节点，2台作为wroker节点。在3台虚拟机上同时部署zookeeper集群，经测试，zookeeper集群联通没问题，其中10.200.33.5是leader节点。具体架构可参加附件图片。
运行docker时，采用net=host模式。
单机训练测试通过。联机时，提示“Fail to initialize ZK connection”，具体见附件图片
求各位大神支招。
Contributor guide
No contributing guide indexed for this repository
Research direction
No source file, test, or entry point is identified. Start by examining the distributed-training code that initializes the ZooKeeper connection and reproduce the failure with the three Docker containers using host networking. Done means the multi-machine training setup initializes ZooKeeper successfully, with the relevant error cause documented.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- cpp, docker
- Domain
- devops, distributed-systems, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 20/100