OpenFaas从入门到实战 – 踩坑指南 ｜ k3d+OpenFaas | deploy your first python function
https://blog.alexellis.io/first-faas-python-function/
https://docs.openfaas.com/deployment/kubernetes/
搭建环境：第一种方法失败，第二种方法亲测有效嘻嘻嘻，其实我大概知道原因，但先不细究了
1. VM: K3s + OpenFaas on Mac
参考教程：https://midnightprogrammer.net/post/installing-openfaas-on-k3s-single-node/
MacBook Pro
使用 Multipass 来创建一个 VM：需要 4GB 内存和 8GB 磁盘，记得要分配多一点 - 我暂定的
beatles@biantongshusMBP ~ % multipass launch --name k3s --memory 4G --disk 8G
Launched: k3s
等待 VM 创建，然后为 VM 启动一个 shell
multipass shell k3s
安装 k3s
curl -sfL https://get.k3s.io | sh -
After the installation is completed, you can check if the k3s service is running by executing the below command.
sudo systemctl status k3s
亲测无效，遂脚本自动安装：
curl -SLsf https://dl.get.arkade.dev/ | sudo sh
谁说用着个很简单，总之我还自己凑了两个命名空间
arkade install openfaas
然后就是这样
Release "openfaas" has been upgraded. Happy Helming!
NAME: openfaas
LAST DEPLOYED: Thu Jan 4 19:51:37 2024
NAMESPACE: openfaas
STATUS: deployed
REVISION: 2
TEST SUITE: None
NOTES:
To verify that openfaas has started, run:
kubectl -n openfaas get deployments -l "release=openfaas, app=openfaas"
To retrieve the admin password, run:
echo $(kubectl -n openfaas get secret basic-auth -o jsonpath="{.data.basic-auth-password}" | base64 --decode)
2024/01/04 19:51:42 stderr: WARNING: Kubernetes configuration file is group-readable. This is insecure. Location: /home/ubuntu/.kube/config
WARNING: Kubernetes configuration file is world-readable. This is insecure. Location: /home/ubuntu/.kube/config
=======================================================================
= OpenFaaS has been installed. =
=======================================================================
# Get the faas-cli
curl -SLsf https://cli.openfaas.com | sudo sh
# Forward the gateway to your machine
kubectl rollout status -n openfaas deploy/gateway
kubectl port-forward -n openfaas svc/gateway 8080:8080 &
# If basic auth is enabled, you can now log into your gateway:
PASSWORD=$(kubectl get secret -n openfaas basic-auth -o jsonpath="{.data.basic-auth-password}" | base64 --decode; echo)
echo -n $PASSWORD | faas-cli login --username admin --password-stdin
faas-cli store deploy figlet
faas-cli list
# For Raspberry Pi
faas-cli store list \
--platform armhf
faas-cli store deploy figlet \
--platform armhf
# Find out more at:
# https://github.com/openfaas/faas
???? Speed up GitHub Actions/GitLab CI + reduce costs: https://actuated.dev
check the rollout status of the gateway, After this we can forward the gateway to the machine.
ubuntu@k3s:/etc/rancher/k3s$ kubectl get pod -n openfaas
NAME READY STATUS RESTARTS AGE
nats-5c48bc8b46-zksff 1/1 Running 2 (12m ago) 73m
alertmanager-795bbdc56c-6qwpn 1/1 Running 2 (12m ago) 73m
prometheus-78d4c9f748-smjfr 1/1 Running 2 (12m ago) 73m
queue-worker-b9965cc56-bn47b 1/1 Running 6 (12m ago) 73m
gateway-67df8c4d4-jfkbf 2/2 Running 0 73m
ubuntu@k3s:/etc/rancher/k3s$ kubectl rollout status -n openfaas deploy/gateway
deployment "gateway" successfully rolled out
ubuntu@k3s:/etc/rancher/k3s$ kubectl port-forward -n openfaas svc/gateway 8080:8080 &
[1] 4737
ubuntu@k3s:/etc/rancher/k3s$ Forwarding from 127.0.0.1:8080 -> 8080
Forwarding from [::1]:8080 -> 8080
jobs
[1]+ Running kubectl port-forward -n openfaas svc/gateway 8080:8080 &
ubuntu@k3s:/etc/rancher/k3s$
You can then view the password by printing the value of the PASSWORD variable using the echo command. 反正报错需要等一会儿就好了
ubuntu@k3s:/etc/rancher/k3s$ echo $PASSWORD
oJPdijCZF2Dd
192.168.64.1
http://localhost:31112/ui
http://127.0.0.1:31112/ui
2. Docker: K3d + OpenFaas
参考教程（版本较老，于是我自己做了一些记录）：https://mickey.dev/posts/getting-started-with-openfaas/#:~:text=Configure%20faas%2Dcli%20%26%20Login%20to%20the%20OpenFaaS%20Dashboard&text=Open%20a%20browser%20and%20navigate,for%20a%20username%20and%20password.
brew install k3d
Create a cluster named mycluster with just a single server node
k3d cluster create mycluster
You can now use it like this:
kubectl cluster-info
事实上，这句话并没有什么用
export KUBECONFIG="$(k3d kubeconfig get --name='k3s-default')"
export KUBECONFIG="$(k3d kubeconfig get --all)"
To install OpenFaaS to your k3d cluster, start by cloning the https://github.com/openfaas/faas-netesrepo:
git clone https://github.com/openfaas/faas-netes.git
cd faas-netes
Install the namespaces:
kubectl apply -f namespaces.yml
This will create two namespaces in your cluster:
openfaas - Which will hold the OpenFaaS cluster services (AKA. ‘Control Plane”).
openfaas-fn - Stores the functions you deploy to OpenFaaS.
Create a password for the OpenFaaS Gateway and add it as a secret into the cluster. The below commands will generate a random password and add the secret:
beatles@biantongshus-MacBook-Pro faas-netes % export PASSWORD=$(head -c 12 /dev/urandom | shasum| cut -d' ' -f1)
beatles@biantongshus-MacBook-Pro faas-netes % kubectl -n openfaas create secret generic basic-auth \
--from-literal=basic-auth-user=admin \
--from-literal=basic-auth-password="$PASSWORD"
secret/basic-auth created
beatles@biantongshus-MacBook-Pro faas-netes %
Now deploy the OpenFaaS stack:
kubectl apply -f ./yaml
Install the faas-cli
The next step is installing the faas-cli. If you’re on MacOS and already have homebrew install then installation is as simple as:
brew install faas-cli
Configure faas-cli & Login to the OpenFaaS Dashboard
Configure the faas-cli to use your local OpenFaaS cluster by using the faas-cli login command. If running k3d you’ll need to forward the gateway service port and set the OPENFAAS_URL environment variable:
kubectl port-forward svc/gateway -n openfaas 31112:8080 &
export OPENFAAS_URL=http://127.0.0.1:31112
echo $PASSWORD | faas-cli login --password-stdin
Open a browser and navigate to http://localhost:31112 to load the UI. The UI will prompt for a username and password. The default username is admin and the password is the one you specified in the deployment instructions above, you can print this to your console using echo $PASSWORD.
echo $PASSWORD
681d0e42c1506c260eba6cc73ad61b3330276302
Cheers!!
再次登陆
http://localhost:31112/
export PASSWORD=681d0e42c1506c260eba6cc73ad61b3330276302
别忘了启动docker engine
kubectl port-forward svc/gateway -n openfaas 31112:8080 &
export OPENFAAS_URL=http://127.0.0.1:31112
echo $PASSWORD | faas-cli login --password-stdin
就好了
3. Deploy New Functions
3.1 bug一览 ｜ 排错完整过程（发疯中）
faas-cli template store list
Error while getting templates info: error while requesting template list: Get "https://raw.githubusercontent.com/openfaas/store/master/templates.json": dial tcp [::]:443: connect: connection refused
祈祷：http://grayblog.cn/2020/07/18/解决Mac中Terminal无法访问github/
IP Lookup : 140.82.113.3 (github.com)
sudo vi /etc/hosts
140.82.113.3 www.github.com
他的ipaddress有好多，我就放了第一个https://www.ipaddress.com/ip-lookup
185.199.109.133 raw.githubusercontent.com
终于！！！不行的话再来一遍
beatles@biantongshusMBP ~ % faas-cli template store list
NAME RECOMMENDED DESCRIPTION SOURCE
bash-streaming [x] openfaas-incubator Bash Streaming template
dockerfile [x] openfaas Classic Dockerfile template
golang-middleware [x] openfaas HTTP middleware interface in Go
java11-vert-x [x] openfaas Java 11 Vert.x template
node18 [x] openfaas HTTP-based Node 18 template
php8 [x] openfaas Classic PHP 8 template
python3-http [x] openfaas Python 3 with Flask and HTTP
python3-http-debian [x] openfaas Python 3 with Flask and HTTP based on Debian
ruby-http [x] openfaas Ruby 2.4 HTTP template
cobol [ ] devries COBOL Template
crystal [ ] tpei Crystal template
crystal-http [ ] koffeinfrei Crystal HTTP template
csharp-httprequest [ ] distantcam C# HTTP template
csharp-kestrel [ ] burtonr C# Kestrel HTTP template
lua53 [ ] affix Lua 5.3 Template
perl-alpine [ ] tmiklas Perl language template based on Alpine image
quarkus-native [ ] pmlopes Quarkus.io native image template
rust [ ] openfaas-incubator Community Rust template
rust-http [ ] openfaas-incubator Community Rust template with HTTP bindings
swift [ ] affix Swift 4.2 Template
vala [ ] affix Vala Template
vala-http [ ] affix Non-Forking Vala Template
vertx-native [ ] pmlopes Eclipse Vert.x native image template
bun-express [ ] openfaas HTTP-based template using bun
csharp [ ] openfaas Classic C# template
go [ ] openfaas Legacy Golang template
golang-http [ ] openfaas Request/response style HTTP template
java11 [ ] openfaas Java 11 template
node [ ] openfaas Legacy Node 12 template
node12 [ ] openfaas HTTP-based Node 12 template
node14 [ ] openfaas HTTP-based Node 14 template
node16 [ ] openfaas HTTP-based Node 16 template
node17 [ ] openfaas HTTP-based Node 17 template
php7 [ ] openfaas Classic PHP 7 template
powershell-http-template [ ] openfaas-incubator Powershell Core HTTP Ubuntu:16.04 template
powershell-template [ ] openfaas-incubator Powershell Core Ubuntu:16.04 template
puppeteer-nodelts [ ] alexellis A puppeteer template for headless Chrome
python [ ] openfaas Classic Python 2.7 template
python27-flask [ ] openfaas Python 2.7 Flask template
python3 [ ] openfaas Classic Python 3 template
python3-debian [ ] openfaas Python 3 Debian template
python3-flask [ ] openfaas Python 3 Flask template
python3-flask-debian [ ] openfaas Python 3 Flask template based on Debian
ruby [ ] openfaas Classic Ruby 2.5 template
faas-cli up --gateway=http://localhost:31112
或
faas-cli build -f ./cpu.yml
faas-cli deploy -f ./cpu.yml --gateway=http://localhost:31112
当然了，探索出过程很痛苦，我把我提的github issue挂在这里 https://github.com/openfaas/faas/issues/1831
事到如今，踩坑真得靠自己了
[0] < Building float-operation done in 104.36s.
[0] Worker done.
Total build time: 104.36s
Deploying: float-operation.
Is OpenFaaS deployed? Do you need to specify the --gateway flag?
Put "http://127.0.0.1:8080/system/functions": dial tcp 127.0.0.1:8080: connect: connection refused
Function 'float-operation' failed to deploy with status code: 500
beatles@biantongshusMBP openfaas % faas-cli logs float-operation
Cannot connect to OpenFaaS on URL: http://127.0.0.1:8080
排错：
kubectl get service -n openfaas
faas-cli deploy -f ./cpu.yml --gateway=http://localhost:31112
最终成功
beatles@biantongshusMBP openfaas % faas-cli login --username admin --password=681d0e42c1506c260eba6cc73ad61b3330276302 --gateway=http://localhost:31112
WARNING! Using --password is insecure, consider using: cat ~/faas_pass.txt | faas-cli login -u user --password-stdin
Calling the OpenFaaS server to validate the credentials...
credentials saved for admin http://localhost:31112
beatles@biantongshusMBP openfaas % faas-cli deploy -f ./cpu.yml --gateway=http://localhost:31112
Deploying: float-operation.
Deployed. 202 Accepted.
URL: http://localhost:31112/function/float-operation
beatles@biantongshusMBP openfaas %
然而遇到not ready
Not Ready
beatles@biantongshusMBP openfaas % kubectl get deploy -n openfaas-fn
NAME READY UP-TO-DATE AVAILABLE AGE
float-operation 0/1 1 0 5h36m
beatles@biantongshusMBP openfaas % kubectl describe -n openfaas-fn deploy/float-operation
Name: float-operation
Namespace: openfaas-fn
CreationTimestamp: Thu, 25 Jan 2024 12:34:41 +0800
Labels: faas_function=float-operation
Annotations: deployment.kubernetes.io/revision: 1
prometheus.io.scrape: false
Selector: faas_function=float-operation
Replicas: 1 desired | 1 updated | 1 total | 0 available | 1 unavailable
StrategyType: RollingUpdate
MinReadySeconds: 0
RollingUpdateStrategy: 0 max unavailable, 1 max surge
Pod Template:
Labels: faas_function=float-operation
Annotations: prometheus.io.scrape: false
Containers:
float-operation:
Image: float-operation:latest
Port: 8080/TCP
Host Port: 0/TCP
Liveness: http-get http://:8080/_/health delay=2s timeout=1s period=2s #success=1 #failure=3
Readiness: http-get http://:8080/_/health delay=2s timeout=1s period=2s #success=1 #failure=3
Environment:
fprocess: python index.py
Mounts: <none>
Volumes: <none>
Conditions:
Type Status Reason
---- ------ ------
Available False MinimumReplicasUnavailable
Progressing False ProgressDeadlineExceeded
OldReplicaSets: <none>
NewReplicaSet: float-operation-bfd748bd6 (1/1 replicas created)
Events: <none>
beatles@biantongshusMBP openfaas %
排错过程：
beatles@biantongshusMBP openfaas % kubectl get deploy -n openfaas-fn
NAME READY UP-TO-DATE AVAILABLE AGE
float-operation 0/1 1 0 5h36m
beatles@biantongshusMBP openfaas % kubectl logs -n openfaas-fn deploy/float-operation
Error from server (BadRequest): container "float-operation" in pod "float-operation-bfd748bd6-4r67p" is waiting to start: trying and failing to pull image
beatles@biantongshusMBP openfaas % kubectl get events -n openfaas-fn \
--sort-by=.metadata.creationTimestamp
LAST SEEN TYPE REASON OBJECT MESSAGE
31m Normal Pulling pod/float-operation-bfd748bd6-p885q Pulling image "float-operation:latest"
37m Warning Failed pod/float-operation-bfd748bd6-p885q Failed to pull image "float-operation:latest": rpc error: code = Unknown desc = failed to pull and unpack image "docker.io/library/float-operation:latest": failed to resolve reference "docker.io/library/float-operation:latest": pull access denied, repository does not exist or may require authorization: server message: insufficient_scope: authorization failed
37m Warning Failed pod/float-operation-bfd748bd6-p885q Error: ErrImagePull
3m28s Normal BackOff pod/float-operation-bfd748bd6-p885q Back-off pulling image "float-operation:latest"
beatles@biantongshusMBP ~ % docker pull float-operation:latest
Error response from daemon: pull access denied for float-operation, repository does not exist or may require 'docker login': denied: requested access to the resource is denied
https://docs.openfaas.com/deployment/kubernetes/
于是我准备先学一下docker: https://www.youtube.com/watch?v=pg19Z8LL06w
现在我学完了，我又来解决问题来了
3.2 First Python Function
faas-cli template store pull python3-http
faas-cli new float-operation --lang python3-http
部署函数
faas-cli up --gateway=http://localhost:31112
或如下几步
faas-cli build -f ./cpu.yml
build失败可能是网络问题 更换节点可以解决
Here’s how to upload the function to a remote registry (if needed):
docker login -u beatlesbian -p Bts210717! docker.io
faas-cli push -f ./cpu.yml
faas-cli deploy -f ./cpu.yml --gateway=http://localhost:31112
出现了这个问题，找到了 https://skemman.is/bitstream/1946/44391/2/OperatingManual.pdf
beatles@biantongshusMBP openfaas % faas-cli push -f ./cpu.yml
Unable to push one or more of your functions to Docker Hub:
- float-operation
You must provide a username or registry prefix to the Function's image such as user1/function1
我需要修改yaml文件，啊！我才懂Alex是什么意思image: beatlesbian/float-operation:latest
终于Push成功
latest: digest: sha256:4f6429e67aea67b421804366a4c89ff39be1dc705cd9a58807fcf656399b9cc2 size: 4695
[0] < Pushing float-operation [beatlesbian/float-operation:latest] done.
[0] Worker done.
然而不是200 ok
beatles@biantongshusMBP openfaas % faas-cli deploy -f ./cpu.yml --gateway=http://localhost:31112
Deploying: float-operation.
Deployed. 202 Accepted.
URL: http://localhost:31112/function/float-operation
这回终于ready了，然而，我谢谢嘞，我能把所有坑踩遍是吧，我怀疑是 an error in the application，毕竟这也是我第一次来写OpenFaas函数，好吧让我去刷牙洗脸，曙光就在前方，我预感明天我能把所有函数都部署完，然后再考虑auto-scaling的问题，然后上hey插件测试性能，年前就做这些吧，每天保证两小时改论文，画一些简单轻松的小画就可以了
The server encountered an internal error and was unable to complete your request. Either the server is overloaded or there is an error in the application.
这是我修改前的 hanlder.py,这种直接返回估计肯定是不对的
import math
from time import time
def float_operations(n):
start = time()
for i in range(0, n):
sin_i = math.sin(i)
cos_i = math.cos(i)
sqrt_i = math.sqrt(i)
latency = time() - start
return latency
def handle(event, context):
n = int(event['n'])
result = float_operations(n)
print(result)
return result
# return {
# "statusCode": 200,
# "body": "Hello from OpenFaaS!"
# }
于是我改成了gpt给我写的，还是不行
这当然就要查看日志了！Your function is crashing due to an error in your code, check the logs. 感谢troubleshooting手册，我们可以很轻松的猜测问题定位在最后一个，当然只是猜测
beatles@biantongshusMBP openfaas % kubectl logs -n openfaas-fn deploy/float-operation
2024/02/07 02:23:31 Version: 0.9.15 SHA: bb7b23f61a2251aa158dac6d409f2fdff383f4f9
2024/02/07 02:23:31 Forking: python, arguments: [index.py]
2024/02/07 02:23:31 Started logging: stderr from function.
.....
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2024/02/07 02:24:48 stderr: File "/home/app/index.py", line 66, in call_handler
2024/02/07 02:24:48 stderr: response_data = handler.handle(event, context)
2024/02/07 02:24:48 stderr: ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2024/02/07 02:24:48 stderr: File "/home/app/function/handler.py", line 16, in handle
2024/02/07 02:24:48 stderr: n = int(event['n'])
2024/02/07 02:24:48 stderr: ~~~~~^^^^^
2024/02/07 02:24:48 stderr: TypeError: 'Event' object is not subscriptable
2024/02/07 02:24:48 POST / - 500 INTERNAL SERVER ERROR - ContentLength: 265B (0.0043s)
beatles@biantongshusMBP openfaas %
The function handler is passed two arguments, event and context.
-
event contains data about the request, including: - body - headers - method - query - path
-
context contains basic information about the function, including: - hostname
问题就出现在event上
n = int(event.body)
I want to test my function without deploying it，尝试一下吧！Then invoke your function via http://127.0.0.1:8081
faas-cli build -f ./cpu.yml
docker run -p 8081:8080 \
--rm \
--name float-operation \
-ti beatlesbian/float-operation:latest
终于成功了！
至此我认为我已经踩完大部分坑了。
3.3 Add Second Function
faas-cli new pyaes --lang python3-http --append cpu.yml
2024/02/07 08:38:07 stderr: import numpy as np
2024/02/07 08:38:07 stderr: ModuleNotFoundError: No module named 'numpy'
缺包可能需要修改dockerfile, 添加RUN ....，再手动build
docker build -t beatlesbian/pyaes:latest ./build/pyaes
然后继续，成功了
faas-cli push -f ./cpu.yml
faas-cli deploy -f ./cpu.yml --gateway=http://localhost:31112
排错常用
kubectl logs -n openfaas-fn deploy/pyaes
对于缺包我做了如下补充：PIL
RUN pip install Pillow
RUN pip install requests
https://blog.csdn.net/qq_28304687/article/details/76551196
根据论文，我现在要把这些都添加完，明天测试openFaas可扩展性，并安装插件
pyaes
RUN pip install pyaes
# test
{
'length_of_message': 4,
'num_of_iterations': 5
}
event_body = json.loads(event.body.decode("utf-8"))
length_of_message = event_body['length_of_message']
num_of_iterations = event_body['num_of_iterations']
{'message': '69wc', 'ciphertext': b'\xb6 \xbf"', 'latency': 0.0022983551025390625}
- TypeError: Object of type bytes is not JSON serializable OpenFaas简单的事例，并没有演示出来event.body是字节码类型，还要靠自己反复摸索，解决方案最终如上，好在可以举一反三了
FunctionBench workloads
1. CPU & Memory
-
⭐️Float Operations(sin, cos, sqrt) done
-
⭐️MatMul(square matrix multiplication) done
-
⭐️Linpack(solve linear equations Ax = b) done
-
⭐️Image Processing
-
⭐️Video Processing
-
MapReduce
-
Chameleon
-
⭐️pyaes done
-
Feature Generation
-
⭐️Model Training
-
Model Serving
- Video Face Detection - Cascade Classifier
- Classification Image - CNN
- Generating Names- RNN
- Prediction Reviews - LR
2. Disk
- ⭐️dd
- ⭐️gzip-compression
3. Network
- iPerf3
- Cloud storage service download-upload
- json serialization deserialization
# from google.cloud import storage
from PIL import Image, ImageFilter
from time import time
import json
import os
TMP = "/tmp/"
#输入的image, file_name, 输出的filename需要斟酌
# left-right + top-bottom
def flip(image, file_name):
path_list = []
path = TMP + "flip-left-right-" + file_name
img = image.transpose(Image.FLIP_LEFT_RIGHT)
img.save(path)
path_list.append(path)
path = TMP + "flip-top-bottom-" + file_name
img = image.transpose(Image.FLIP_TOP_BOTTOM)
img.save(path)
path_list.append(path)
return path_list
# rotate 90 180 270
def rotate(image, file_name):
path_list = []
path = TMP + "rotate-90-" + file_name
img = image.transpose(Image.ROTATE_90)
img.save(path)
path_list.append(path)
path = TMP + "rotate-180-" + file_name
img = image.transpose(Image.ROTATE_180)
img.save(path)
path_list.append(path)
path = TMP + "rotate-270-" + file_name
img = image.transpose(Image.ROTATE_270)
img.save(path)
path_list.append(path)
return path_list
# filter
def filter(image, file_name):
path_list = []
path = TMP + "blur-" + file_name
img = image.filter(ImageFilter.BLUR)
img.save(path)
path_list.append(path)
path = TMP + "contour-" + file_name
img = image.filter(ImageFilter.CONTOUR)
img.save(path)
path_list.append(path)
path = TMP + "sharpen-" + file_name
img = image.filter(ImageFilter.SHARPEN)
img.save(path)
path_list.append(path)
return path_list
def gray_scale(image, file_name):
path = TMP + "gray-scale-" + file_name
img = image.convert('L')
img.save(path)
return [path]
def resize(image, file_name):
path = TMP + "resized-" + file_name
image.thumbnail((128, 128))
image.save(path)
return [path]
# upon 5 image processing actions
def image_processing(file_name, image_path):
path_list = []
start = time()
with Image.open(image_path) as image:
tmp = image
path_list += flip(image, file_name)
path_list += rotate(image, file_name)
path_list += filter(image, file_name)
path_list += gray_scale(image, file_name)
path_list += resize(image, file_name)
latency = time() - start
result = {
'path_list': path_list,
'latency': latency
}
return result
# file_name, image_path
def handle(event, context):
image_path = str(event.body)
file_name = os.path.basename(image_path)
result = image_processing(file_name,image_path)
return {
"statusCode": 200,
"body": json.dumps(result)
}
Kubeedge
multipass launch --name kubeedge1 --memory 4G --disk 8G