anaconda环境---ubuntu下重装
@wp20190312
为何重装？ 配置一个环境，意外发现conda命令不好用了，提示“找不到conda模块”，整个conda虚拟环境中的工程项目无法使用，网上有说主要是python路径的指向调用出现了问题。经过多番尝试仍然未解决问题，所以还是重装一下吧，简单bao力。
原来使用的是anaconda2，这次改用anaconda3。
（1）卸载原来的anaconda2：rm -rf /home/wp/anaconda2
（2）安装新的anaconda3
在网址 https://www.cnblogs.com/gaofighting/p/8799169.html 中下载anaconda3
wp@wp-MS-7519:~/Anaconda3$ sudo sh Anaconda3-2018.12-Linux-x86_64.sh #安装完后带锁
sh Anaconda3-2018.12-Linux-x86_64.sh #安装完后不带锁
=======================安装过程部分展示================================
一直enter到选择yes?
Anaconda3 will now be installed into this location:
/home/wp/anaconda3
- Press ENTER to confirm the location
- Press CTRL-C to abort the installation
- Or specify a different location below
[/home/wp/anaconda3] >>>
PREFIX=/home/wp/anaconda3
# 安装根据机器性能，通常需要持续几分钟。
Do you wish the installer to initialize Anaconda3
in your /home/wp/.bashrc ? [yes|no]
[no] >>>
You may wish to edit your /home/wp/.bashrc to setup Anaconda3:
source /home/wp/anaconda3/etc/profile.d/conda.sh
Thank you for installing Anaconda3!
====================================================================
To install Visual Studio Code, you will need:
- Administrator Privileges
- Internet connectivity
Visual Studio Code License: https://code.visualstudio.com/license
Do you wish to proceed with the installation of Microsoft VSCode? [yes|no]
>>> yes
Proceeding with installation of Microsoft VSCode
VSCode is already installed!
到此Anaconda3安装成功。 重启终端，即可使用Anaconda3
假如，Anaconda3未安装成功，可以执行: rm -rf ~/anaconda3 ，找到Anaconda3的安装路径，卸载后重新安装即可。
===============================================================
（3）在终端，查看当前电脑的python版本：
wp@wp-MS-7519:~/anaconda3$ python
Python 2.7.15rc1 (default, Apr 15 2018, 21:51:34)
[GCC 7.3.0] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> exit()
wp@wp-MS-7519:~/anaconda3$ python3
Python 3.6.5 (default, Apr 1 2018, 05:46:30)
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> exit()
我们可以看到，若在终端输入 python，仍然会显示Ubuntu自带的python版本。
（4）所以如果想用anaconda，则需要修改终端的默认python为anaconda，执行：
wp@wp-MS-7519:~/anaconda3$ gedit ~/.bashrc
在文件后面添加：export PATH="/home/wp/anaconda3/bin:$PATH" #并保存
wp@wp-MS-7519:~/anaconda3$ source ~/.bashrc #使用修改的路径生效
【export PATH=/home/wp/anaconda2/bin:$PATH
gedit ~/.bashrc
source ~/.bashrc】
（5）再次，查看当前电脑的python版本：
wp@wp-MS-7519:~/anaconda3$ python -V
Python 3.7.1 #anaconda3 安装的python版本
wp@wp-MS-7519:~/anaconda3$ python
Python 3.7.1 (default, Dec 14 2018, 19:28:38)
[GCC 7.3.0] :: Anaconda, Inc. on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> exit()
wp@wp-MS-7519:~/anaconda3$ python2
Python 2.7.15rc1 (default, Apr 15 2018, 21:51:34)
[GCC 7.3.0] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>>
（6）根据工作要求，增加虚拟环境。
6.1查找可用版本：conda search "^python$"
wp@wp-MS-7519:~/anaconda3$ conda search "^python$"
Loading channels: done
# Name Version Build Channel
python 1.0.1 0 anaconda/cloud/conda-forge
python 1.0.1 0 anaconda/pkgs/free
........
python 2.7.3 2 anaconda/pkgs/free
python 2.7.3 3 anaconda/pkgs/free
python 2.7.3 4 anaconda/pkgs/free
python 2.7.3 5 anaconda/pkgs/free
......................
python 2.7.15 0 anaconda/cloud/conda-forge
python 2.7.15 h1571d57_0 anaconda/pkgs/main
python 2.7.15 h33da82c_1 anaconda/cloud/conda-forge
python 2.7.15 h33da82c_3 anaconda/cloud/conda-forge
python 2.7.15 h33da82c_4 anaconda/cloud/conda-forge
...
python 2.7.15 h9bab390_2 anaconda/pkgs/main
python 2.7.15 h9bab390_4 anaconda/pkgs/main
python 2.7.15 h9bab390_6 anaconda/pkgs/main
python 2.7.15 h9fef7bc_0 anaconda/cloud/conda-forge
................................
python 3.5.0 0 anaconda/pkgs/free
python 3.5.0 1 anaconda/pkgs/free
python 3.5.1 0 anaconda/cloud/conda-forge
python 3.5.1 0 anaconda/pkgs/free
python 3.5.1 1 anaconda/cloud/conda-forge
python 3.5.1 5 anaconda/pkgs/free
python 3.5.2 0 anaconda/cloud/conda-forge
python 3.5.2 0 anaconda/pkgs/free
python 3.5.2 1 anaconda/cloud/conda-forge
...
python 3.5.5 hc3d631a_0 anaconda/pkgs/main
python 3.5.5 hc3d631a_1 anaconda/pkgs/main
python 3.5.5 hc3d631a_3 anaconda/pkgs/main
python 3.5.5 hc3d631a_4 anaconda/pkgs/main
python 3.5.6 hc3d631a_0 anaconda/pkgs/main
python 3.6.0a3 0 anaconda/cloud/conda-forge
............................
python 3.6.7 hd21baee_1002 anaconda/cloud/conda-forge
python 3.6.8 h0371630_0 anaconda/pkgs/main
python 3.7.0 h5001a0f_0 anaconda/cloud/conda-forge
python 3.7.0 h5001a0f_1 anaconda/cloud/conda-forge
python 3.7.0 h5001a0f_4 anaconda/cloud/conda-forge
python 3.7.0 h5001a0f_6 anaconda/cloud/conda-forge
python 3.7.0 h6e4f718_3 anaconda/pkgs/main
...
python 3.7.1 h0371630_7 anaconda/pkgs/main
python 3.7.1 h381d211_1002 anaconda/cloud/conda-forge
python 3.7.1 h5001a0f_0 anaconda/cloud/conda-forge
python 3.7.1 hd21baee_1000 anaconda/cloud/conda-forge
python 3.7.1 hd21baee_1001 anaconda/cloud/conda-forge
python 3.7.2 h0371630_0 anaconda/pkgs/main
6.2下载的TensorFlow对应的Python版本一定要和conda create -n tensorflow python=x.x的版本一样才行。
考虑，环境没破坏之前，FaceNet使用的环境：Ubuntu 18.04 + Tensorflow 1.5.0 + Python 2.7 + OpenCV 3.2.0，
A1，这里，装一个Python 2.7，配一个Tensorflow 1.5.0 环境。
wp@wp-MS-7519:~/anaconda3$ conda create -n test_py2 python=2
wp@wp-MS-7519:~/anaconda3$ source activate test_py2
wp@wp-MS-7519:~/anaconda3$ python #查看是python2.7.15
wp@wp-MS-7519:~/anaconda3$ pip install tensorflow
DEPRECATION: Python 2.7 will reach the end of its life on January 1st, 2020. Please upgrade your Python as Python 2.7 won't be maintained after that date. A future version of pip will drop support for Python 2.7.
Looking in indexes: http://pypi.douban.com/simple
Collecting tensorflow
Downloading http://pypi.doubanio.com/packages/d2/ea/ab2c8c0e81bd051cc1180b104c75a865ab0fc66c89be992c4b20bbf6d624/tensorflow-1.13.1-cp27-cp27mu-manylinux1_x86_64.whl (92.5MB)
B1，安装完成后，python ,import tensorflow as tf,报错
>>> import tensorflow as tf
2019-03-11 16:39:04.749263: F tensorflow/core/platform/cpu_feature_guard.cc:37] The TensorFlow library was compiled to use SSE4.2 instructions, but these aren't available on your machine.
已放弃 (核心已转储)
解决办法:安装bazel，编译tensorflow。这个问题的原因是CPU支持更快的运算，安装的Tensorflow中缺少对应的模块，需要编译安装。当我们使用GPU的情况下，并不需要用到CPU，所以这些warning可以忽略。 安装bazel 太麻烦了，换思路解决“安装tf”。
C1，卸载tensorflow：(test_py2) wp@wp-MS-7519:~/anaconda3$ pip uninstall tensorflow
D1，重新安装tensorflow：sudo pip install --ignore-installed --upgrade https://mirrors.tuna.tsinghua.edu.cn/tensorflow/linux/cpu/tensorflow-1.5.0-cp27-none-linux_x86_64.whl
这里说明一下: 这一步安装之后，python2下面import tensorflow as tf 还是报错，但是在python3下面import tensorflow as tf 不报错。所以，换思路。
A2，换思路，装一个Python 3.5，配一个Tensorflow 1.5.0 环境。【刚开始装的是Python 3.6和Tensorflow 1.5.0 最后不好用，经过折腾换Python 3.5好用。】
wp@wp-MS-7519:~/anaconda3$ conda create -n test_py3 python=3.6
wp@wp-MS-7519:~/anaconda3$ source activate test_py3
(test_py3) wp@wp-MS-7519:~/anaconda3$ cd test_py3
(test_py3) wp@wp-MS-7519:~/anaconda3/test_py3$ cd tf
(test_py3) wp@wp-MS-7519:~/anaconda3/test_py3/tf$ python
Python 3.6.7 | packaged by conda-forge | (default, Feb 28 2019, 09:07:38)
[GCC 7.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>>
安装tf：sudo pip install --ignore-installed --upgrade https://mirrors.tuna.tsinghua.edu.cn/tensorflow/linux/cpu/tensorflow-1.5.0rc0-cp36-cp36m-linux_x86_64.whl
安装tf报错：tensorflow-1.5.0rc0-cp36-cp36m-linux_x86_64.whl is not a supported wheel on this platform.
退出虚拟环境，并卸载python3.6,重装python3.5
(test_py3) wp@wp-MS-7519:~/anaconda3/test_py3/tf$ source deactivate （conda deactivate） #退出虚拟环境
wp@wp-MS-7519:~/anaconda3/test_py3/tf$ conda remove --name test_py3 --all #卸载python3.6
wp@wp-MS-7519:~/anaconda3/test_py3/tf$ conda create -n test_py3 python=3.5 #重装python3.5
(test_py3) wp@wp-MS-7519:~/anaconda3/test_py3/tf$ conda activate test_py3
(test_py3) wp@wp-MS-7519:~/anaconda3/test_py3/tf$ pip install --ignore-installed --upgrade tensorflow-1.5.0-cp35-cp35m-linux_x86_64.whl
【安装下载镜像网址 https://mirrors.tuna.tsinghua.edu.cn/tensorflow/linux/cpu/】
(test_py3) wp@wp-MS-7519:~/anaconda3/test_py3/tf$ python
Python 3.5.5 | packaged by conda-forge | (default, Jul 23 2018, 23:45:43)
[GCC 4.8.2 20140120 (Red Hat 4.8.2-15)] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import tensorflow as tf
>>> tf.__version__
'1.5.0'
>>>
提示，这里需要 py 和 tf 版本相匹配，所以可能会出现反复的：
（1）安装 conda create -n test_py3 python=3.5 与卸载 conda remove --name test_py3 --all ，
（2）安装 pip install --ignore-installed --upgrade tf_setup_name 与卸载 pip uninstall tensorflow 。
【
总结说明：
安装可用环境：conda create --name my_env python=3
删除环境变量：conda remove --name my_env --all
激活新环境：source activate my_env
停用环境：source deactivate
安装可用环境python3.5版本：conda create -n my_env35 python=3.5
删除可用环境python3.5版本：conda remove --name my_env35 --all
激活新环境：source activate my_env
停用：source deactivate
更新版本：conda update python=3.5.2
查看所有环境：conda info --envs
给环境安装其他软件包：conda install --name my_env35 numpy
更新 anaconda：conda update conda
卸载Anaconda：conda install anaconda-clean --yes
rm -rf ~/anaconda3
vim ~/.bashrc
更换仓库镜像
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/free/
conda config --add channels https://mirrors.tuna.tsinghua.edu.cn/anaconda/pkgs/main/
conda config --set show_channel_urls yes
】