[KAFKA-9621] AdminClient listOffsets operation does not respect retries and backoff

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: admin
• Assignee: Cheng Tan
• Reporter: Sanjana Kaundinya
• Created: 27/Feb/20 21:55
• Updated: 20/Oct/20 17:01

Description

Similar to https://issues.apache.org/jira/browse/KAFKA-9047, currently the listOffsets operation doesn't respect the configured retries and backoff for a given call. 

For example, the code path could go like so:
1) Make a metadata request and schedule subsequent list offsets calls
2) Metadata error comes back with InvalidMetadataException

3) Go back to 1

The problem here is that the state is not preserved across calls. We loose the information regarding how many tries the call has been tried and how far out we should schedule the call to try again. This could lead to a tight retry loop and put pressure on the brokers.

Issue Links

• is related to: KAFKA-9678 Introduce bounded exponential backoff in clients