getIPV4Address failed when using PAIK8S mode

[xuehui1991]

When using the PAIK8S mode, I got some error:

Training service error:
getIPV4Address() fail because os.networkinterfaces().eth0 is undefined.

By the way, I already set the "nniManagerIp" in config correctly.

Short summary about the issue/question:

Brief what process you are following:

How to reproduce it:

nni Environment:

• nni version: 1.3
• nni mode(local|pai|remote): paik8s
• OS: linux
• python version: .36
• is conda or virtualenv used?: conda
• is running in docker?: yes

need to update document(yes/no):

Anything else we need to know:

[SparkSnail]

This error is caused by the python environment. There are two NNI environment in user's machine, and one env is for 'local python', one is for 'virtual environment python', the two environment set nnictl's path to the same place, and the 'local python' nnictl path override the 'virtual environment python' nnictl path. When user use nnictl command, the system will always use 'local python' nnictl, while user want to use 'virtual environment python' nnictl.

[ICEORY]

Hi, I also get the same error: getIPV4Address() fail because os.networkinterfaces().eth0 is undefined; I also test nni 1.2, and it works fine.

log

[2/8/2020, 6:20:18 AM] INFO [ 'NNIManager setClusterMetadata, key: nni_manager_ip, value: {"nniManagerIp":"10.10.1.139"}' ]
[2/8/2020, 6:20:18 AM] ERROR [ 'Uknown key: nni_manager_ip' ]

{ Error: getIPV4Address() failed because os.networkInterfaces().eth0 is undefined.
    at Object.getIPV4Address (/usr/local/nni/common/utils.js:183:15)
    at PAIK8STrainingService.submitTrialJobToPAI (/usr/local/nni/training_service/pai/paiK8S/paiK8STrainingService.js:170:103)
    at process._tickCallback (internal/process/next_tick.js:68:7)
  name: '',
  cause:
   Error: getIPV4Address() failed because os.networkInterfaces().eth0 is undefined.
       at Object.getIPV4Address (/usr/local/nni/common/utils.js:183:15)
       at PAIK8STrainingService.submitTrialJobToPAI (/usr/local/nni/training_service/pai/paiK8S/paiK8STrainingService.js:170:103)
       at process._tickCallback (internal/process/next_tick.js:68:7) }

environment:

• nni 1.3
• python 3.6
• openpai 0.14
• ubuntu 16.04
• running in docker
• without anaconda

[SparkSnail]

Hi @iceroy, this issue is fixed by #1987, will release this fix on v1.4.