Cluster: cluster coworkers update of raw api

[aesirteam]

Implementing automatic discovery of coworkers in the origin cluster in the K8S environment.

Principle:

1. Create a headless service for the origin server through the K8S API.
2. Create a stateful set for the origin server through the K8S API.
3. Use K8S DNS to query the headless service and obtain the IP addresses of the origin server pods.
4. Use the RAW API to update coworkers and achieve elastic scaling for the origin server's horizontal pod autoscaler (HPA).
5. Utilize a livenessProbe to call a shell script for updating coworkers, ensuring the correct lifecycle of the pods while also maintaining health checks.

---

TRANS_BY_GPT3

[codecov-commenter]

Codecov Report

Merging #1887 into develop will decrease coverage by 0.07%.
The diff coverage is 0.00%.

@@             Coverage Diff             @@
##           develop    #1887      +/-   ##
===========================================
- Coverage    55.84%   55.77%   -0.08%     
===========================================
  Files          121      121              
  Lines        50701    50769      +68     
===========================================
- Hits         28315    28314       -1     
- Misses       22386    22455      +69     

Impacted Files	Coverage Δ
trunk/src/app/srs_app_config.cpp	68.08% <0.00%> (-0.20%)	⬇️
trunk/src/app/srs_app_config.hpp	100.00% <ø> (ø)
trunk/src/app/srs_app_coworkers.cpp	0.00% <0.00%> (ø)
trunk/src/app/srs_app_http_api.cpp	0.00% <0.00%> (ø)
trunk/src/app/srs_app_reload.cpp	2.50% <0.00%> (-0.07%)	⬇️
trunk/src/app/srs_app_server.cpp	0.00% <0.00%> (ø)
trunk/src/kernel/srs_kernel_utility.cpp	90.37% <0.00%> (-1.39%)	⬇️
trunk/src/protocol/srs_service_utility.cpp	72.43% <0.00%> (-0.55%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0f62587...efce62d. Read the comment docs.

TRANS_BY_GPT3

[aesirteam]

K8S Deployment File

k8s-srs-clusters.yaml

TRANS_BY_GPT3

[winlinvip]

Thank you very much for submitting this PR and providing a solution to the problem. 👍
Make sure to maintain the markdown structure.

By using the service to discover the origin server and updating the configuration of each origin server through the HTTP RAW API, new instances of the origin server can be discovered. This is a very direct approach and does not rely on other services, which is great.
Make sure to maintain the markdown structure.

The essence of this problem is that when the origin server cluster is scaled, it cannot automatically discover new instances of the origin server because the configuration file of the origin server has fixed IP or domain names. The most direct solution is to dynamically modify the configuration file.
Make sure to maintain the markdown structure.

The problem with this solution lies in the HTTP RAW API, which is very unstable because it requires modifying the configuration and then reloading it. Especially in K8S, configuration files are usually configured and mounted to pods through configMap, and modifying the configMap will automatically trigger a reload.
Make sure to maintain the markdown structure.

The RAW API solution is actually conflicting with the reload configuration solution. RAW requires the configuration to be writable, while configMap is read-only (or it is best not to be written by the program to avoid synchronization issues with multiple writes).
Make sure to maintain the markdown structure.

The best way to scale the origin server cluster is to configure coworkers not with their individual IP addresses, but with the address of a service that manages the origin servers. Please refer to OCM: #1607 (comment).
Make sure to maintain the markdown structure.

TRANS_BY_GPT3