feat: Support APISIX_PROFILE for env-specific configuration

[kevinw66]

Why submit this pull request?

• [*] New feature provided

What changes will this PR take into?

Support env config just like apisix
https://apisix.apache.org/zh/docs/apisix/profile/

Related issues

fix/resolve #2194

[zaunist]

Hi @kevinw66 ,it looks great.
cc @juzhiyuan, please unlock the CI to test this PR.

[codecov-commenter]

Codecov Report

Merging #2238 (289d184) into master (ae1001d) will increase coverage by 4.76%.
The diff coverage is 50.00%.

@@            Coverage Diff             @@
##           master    #2238      +/-   ##
==========================================
+ Coverage   65.08%   69.85%   +4.76%     
==========================================
  Files         184      184              
  Lines        7274     7277       +3     
  Branches      829      829              
==========================================
+ Hits         4734     5083     +349     
+ Misses       2256     1899     -357     
- Partials      284      295      +11     

Flag	Coverage Δ
backend-e2e-test	41.06% <50.00%> (+0.17%)	⬆️
backend-e2e-test-ginkgo	56.65% <50.00%> (?)
backend-unit-test	49.16% <ø> (ø)
frontend-e2e-test	68.06% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
api/internal/conf/conf.go	57.54% <50.00%> (-0.71%)	⬇️
api/internal/core/storage/etcd.go	51.18% <0.00%> (+3.93%)	⬆️
api/internal/core/store/validate.go	70.70% <0.00%> (+4.04%)	⬆️
api/internal/handler/upstream/upstream.go	86.58% <0.00%> (+4.87%)	⬆️
api/internal/core/store/store.go	90.62% <0.00%> (+5.20%)	⬆️
api/internal/filter/authentication.go	78.94% <0.00%> (+5.26%)	⬆️
api/internal/handler/service/service.go	93.54% <0.00%> (+5.64%)	⬆️
api/internal/handler/ssl/ssl.go	74.40% <0.00%> (+5.68%)	⬆️
api/internal/route.go	87.50% <0.00%> (+7.50%)	⬆️
api/internal/handler/route/route.go	80.23% <0.00%> (+7.75%)	⬆️
... and 15 more

---

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 ae1001d...289d184. Read the comment docs.

[juzhiyuan]

Hi @kevinw66, please take a look at the CI 😄

[zaunist]

@juzhiyuan It looks like the CI failed due to unstable CI, I found stream route test untill very unstable, the part of the test is easy to fail.
@bzp2010 Can you take some time to look the stream route test ?

[juzhiyuan]

I just rerun the CI, wait for a moment.

[kevinw66]

I merged new commits from master.
BTW, the ut is hard to run in local machine because of too many dependencies server, is there any plan to solve this? like mock server or embbed server.

[zaunist]

I merged new commits from master. BTW, the ut is hard to run in local machine because of too many dependencies server, is there any plan to solve this? like mock server or embbed server.

We can use make api-test to run unit test, I think you might be talking about integration testing?

[kevinw66]

I merged new commits from master. BTW, the ut is hard to run in local machine because of too many dependencies server, is there any plan to solve this? like mock server or embbed server.

We can use make api-test to run unit test, I think you might be talking about integration testing?

I tried to run failed test stream_route_test in my local machine, but I found I have to start dashboard in local, and then run an upstream, or else it will failed in connection refused.

[juzhiyuan]

Hi @zaunist, could you please give some hints on executing tests locally?

[zaunist]

I also have this problem in my PR submission. @bzp2010 more familiar with this, any solutions?

[bzp2010]

Hi, @zaunist.
Recently we have not modified the E2E test implemented using the old way, and its instability is perhaps caused by the test environment.
And in the new CI using ginkgo, I have solved one instability issue related to migrate, but it still may fail due to temporary environment exceptions. Also, the UDP case in the Stream Route test does still have some failure issues, which in general are not very common.

[bzp2010]

Hi, @kevinw66. It's great to see you contributing to the project.

But do we really need this feature? As far as I know, we can now rewrite the config file path using the -c flag. Can you describe some of the differences between the current PR and -c flag scenarios?

Thanks.

[kevinw66]

Hi, @kevinw66. It's great to see you contributing to the project.

But do we really need this feature? As far as I know, we can now rewrite the config file path using the -c flag. Can you describe some of the differences between the current PR and -c flag scenarios?

Thanks.

When we download bin file and use it directly, -c is enough. But if we have to redevelop from source code, mostly we will create env-specific config, and the -c may not so quite useful.
On the other hand, I think it's better to keep dashboard project and main project have same behavior, like config-default 、environment variable like etcd.host: ${{ETCD_HOST}} and so on.

[bzp2010]

Hi, @kevinw66.

Sorry to reply to your delay, I miss it when dealing with PR, I have learned the purpose of this function, this feature is also used in Apache APISIX, I think it is no problem, but can you add some tests to it?

[bzp2010]

Still need some testing to ensure it works fine. 🤔