feat(csharp/src/Drivers/Databricks): Added connection params of rateLimitRetry and rateLimitRetryTimeout #3612
Conversation
…atusCode)429 to ensure compatibility across all target frameworks
|
possible to create some mock test on this? |
eric-wang-1990
left a comment
eric-wang-1990
left a comment
There was a problem hiding this comment.
It seems for other error we are already using exponential backoff, why do we need special case for 429? Just change the timeout seems not that valuable to me.
Does ODBC/JDBC has this special parameter for 429 retry?
Just wanna understand the context a little bit more.
ODBC has a param for rateLimitRetryTimeout specific to 429. The parameter refers to the number of seconds that the connector waits before stopping an attempt to retry an operation when the operation receives an HTTP 429 response with Retry-After headers. |
5 participants
Summary
Adds support for separate retry configuration for HTTP 429 (Too Many Requests) responses, allowing independent control of rate limit retry behavior from other retryable errors (408, 502, 503, 504).
Changes Made
adbc.databricks.rate_limit_retry(boolean, default: true): Controls whether to retry HTTP 429 (TooManyRequests) responses. Can be disabled independently from other retry typesExample:
"adbc.databricks.rate_limit_retry": "false"adbc.databricks.rate_limit_retry_timeout(integer, default: 120 seconds): Maximum total time in seconds to retry HTTP 429 responses before failing. Set to 0 to retry indefinitelySeparate from the
TemporarilyUnavailableRetryTimeout(900 seconds for 408/502/503/504)Example:
"adbc.databricks.rate_limit_retry_timeout": "60"