Real-time-Anomaly-Detection

Are you crazy? Or the network is crazy?

D1

build container with Dockerfile

cd D1
sudo docker build -t jiahong/lp-jupyter -f Dockerfile .

create and mount container to local host

sudo docker run -dit \
--name jupyter-container \
--mount type=bind,source=/home/jiahong/liveProject/Real-time-Anomaly-Detection,target=/src \
-p 8080:8888 \
jiahong/lp-jupyter

check the token

sudo docker exec -it <container> bash

D2

build container with Dockerfile

cd D2
sudo docker build -t jiahong/lp-jupyter:D2 -f Dockerfile .

create and mount container to local host

sudo docker run -dit \
--name jupyter-container \
--mount type=bind,source=/home/jiahong/liveProject/Real-time-Anomaly-Detection,target=/src \
-p 8080:8888 \
jiahong/lp-jupyter:D2

To show the token, first exec into the container

sudo docker exec -it <container> bash

Then,

jupyter notebook list

D3

use Makefile to build and run docker

Makefile looks like:

run:
    sudo docker build xxx
    sudo docker run xxx

Enter the $pwd and run

make run

Persistant storage of the trained weights

from joblib import dump, load
dump(clf, 'filename.joblib')

load the weights

clf = load('filename.joblib')

Connect 2 matrxi by coloum -> np.c_

a = np.array([[1, 2, 3], [7, 8, 9]])
b = np.arry([[4, 5, 6], [8, 8, 8]])
c = np.c_[a, b]

To draw the decision frontier, enough sample points as the background layer and trained classfier are required. The workflow is first generate dense enough sample points and punch them through the classfier and color them by the classification result.

# set the figure size
plt.rcParams['figure.figsize'] = [15, 15]
# generate dense enough sample points
xx, yy = np.meshgrid(np.linspace(-2, 70, 100), np.linspace(-2, 70, 100))
# punch the sample points through the classfier
Z = clf.decision_function(np.c_[xx.ravel(), yy.ravel()])
Z = Z.reshape(xx.shape)

# 等高线
plt.contourf(xx, yy, Z, levels=np.linspace(Z.min(), 0, 8), cmap=plt.cm.PuBu, alpha=0.5)
plt.contour(xx, yy, Z, levels=[0], linewidths=2, colors='g')

D4

Got fastapi template.

check the current logs in the container

sudo docker logs <container>

D5

Old friend.

Problem: ERROR: Couldn't connect to Docker daemon at http+docker://localhost - is it running?

sudo usermod -aG docker $USER
sudo ln -s /usr/local/bin/docker-compose /usr/bin/docker-compose
sudo service docker restart

Caddy: Panic! casued by Ubutnu 20.04 LTS's resolv.conf. Pass the modified resolv.conf to caddy instance. The modified file is like following:

nameserver 127.0.0.11
options edns0 ndots:0
search internet-only.domain

D6

Rebuild the fastapt imgae with metrics. Update the requirements.txt and rebuid it with makefile:

run:
	sudo docker build -t jiahong/lp-fastapi:D6 -f Dockerfile .

Run the docker-compose

ADMIN_USER=admin ADMIN_PASSWORD=admin docker-compose up -d

Stop the current docker-compose

docker-compose down -v

Cannot access the prometheus in the grafana container. The solution is to replace the ip address in host network by the ip in docker network

docker inspect <prometheus container>

# no http://
192.168.144.3:9090