The primary worker pool is getting exhausted when the endpoint security is configured

[ksankeerth]

Current Limitation

Currently, there is no way to configure socket timeout, connection timeout and request timeout for oauth token endpoints configured for HTTP endpoints. Therefore, If the token endpoint is slow to connect, the passthrough message processor pool can be exhausted.

Reproducing Steps:

• Configure an API with oauth protected endpoint and have some token endpoint delay (eg ~5 seconds)
• Send 500 requests within 1 seconds

Observations

• Primary passthrough message processor pool will be get exhausted

Suggested Improvement

• Allow a way to configure socket timeout , request timeout and connection timeout (or introduce some default values to prevent passthrough message processor pool get exhausted)
• improve current logic to check the token in cache before every new requests is made to to token endpoint

Version

APIM 4.0.0