Set a sensible default for API server HTTP Read buffer size

[gberche-orange]

Pull request description

Set a sensible default for API server HTTP Read buffer size

Avoid explicit use of the following when using oauth2-proxy in front of api server

    testkube-api:
      extraEnvVars:
      - name: APISERVER_HTTP_READBUFFERSIZE
        value: "8192"

Relates to kubeshop/testkube#2870

Checklist (choose whats happened)

• breaking change! (describe)
• tested locally: only tested extraEnvVars workaround so far
• tested on cluster
• added new dependencies
• updated the docs
• added a test

Breaking changes

Changes

Fixes

[vsukhin]

looks good, @ypoplavs ?

[alfeg]

For me 32000 were required to make testkube work under Azure AD auth

ALSO this is quite a breaking change for us.

As it causes testkube fail to update in case if snippet exists in values.yaml file as proposed in one the issues kubeshop/testkube#3141 (comment)

testkube-api:
   extraEnvVars:
   - name: APISERVER_HTTP_READBUFFERSIZE
     value: '32000'

Error: UPGRADE FAILED: cannot patch "testkube-api-server" with kind Deployment: The order in patch list:
[map[name:POSTMANEXECUTOR_URI value:] map[name:SLACK_TOKEN value:] map[name:STORAGE_REGION value:] map[name:STORAGE_TOKEN value:] map[name:JOB_SERVICE_ACCOUNT_NAME value:] map[name:APISERVER_HTTP_READBUFFERSIZE value:8192] map[name:APISERVER_HTTP_READBUFFERSIZE value:32000] map[name:TESTKUBE_OAUTH_CLIENTID value:] map[name:TESTKUBE_OAUTH_CLIENTSECRET value:] map[name:TESTKUBE_OAUTH_SCOPES value:] map[name:TESTKUBE_WATCHER_NAMESPACES value:] map[name:TESTKUBE_REGISTRY value:<nil>] map[name:CDEVENTS_TARGET value:] map[name:TESTKUBE_DASHBOARD_URI value:]]
 doesn't match $setElementOrder list:
[map[name:APISERVER_PORT] map[name:API_MONGO_DSN] map[name:API_MONGO_ALLOW_DISK_USE] map[name:NATS_URI] map[name:POSTMANEXECUTOR_URI] map[name:TESTKUBE_ANALYTICS_ENABLED] map[name:SLACK_TOKEN] map[name:STORAGE_ENDPOINT] map[name:STORAGE_BUCKET] map[name:STORAGE_EXPIRATION] map[name:STORAGE_ACCESSKEYID] map[name:STORAGE_SECRETACCESSKEY] map[name:STORAGE_REGION] map[name:STORAGE_TOKEN] map[name:STORAGE_SSL] map[name:SCRAPPERENABLED] map[name:LOGS_BUCKET] map[name:LOGS_STORAGE] map[name:TESTKUBE_NAMESPACE] map[name:JOB_SERVICE_ACCOUNT_NAME] map[name:APISERVER_FULLNAME] map[name:APISERVER_CONFIG] map[name:APISERVER_HTTP_READBUFFERSIZE] map[name:TESTKUBE_OAUTH_CLIENTID] map[name:TESTKUBE_OAUTH_CLIENTSECRET] map[name:TESTKUBE_OAUTH_PROVIDER] map[name:TESTKUBE_OAUTH_SCOPES] map[name:TESTKUBE_CLOUD_URL] map[name:TESTKUBE_WATCHER_NAMESPACES] map[name:APISERVER_HTTP_READBUFFERSIZE] map[name:TESTKUBE_CONFIG_DIR] map[name:TESTKUBE_REGISTRY] map[name:TESTKUBE_POD_START_TIMEOUT] map[name:CDEVENTS_TARGET] map[name:TESTKUBE_DASHBOARD_URI]]

Solution is to remove APISERVER_HTTP_READBUFFERSIZE extra env and replace it with httpReadBufferSize: 32000 (or whatever value You use)

[vsukhin]

@ypoplavs please check it out

[ypoplavs]

Hello @alfeg!
Thank you for your comment and apologies for the breaking change that caused your TK installation to fail. We will be testing our changes more thoroughly.

I have checked our values.yaml file and saw that we do use httpReadBufferSize value and not APISERVER_HTTP_READBUFFERSIZE, thus for installations with updated values file it will work.