Sergey Soldatov created RATIS-2385:
--------------------------------------
Summary: Don't keep failed requests in the RetryCache.
Key: RATIS-2385
URL: https://issues.apache.org/jira/browse/RATIS-2385
Project: Ratis
Issue Type: Improvement
Components: common
Affects Versions: 3.2.1
Reporter: Sergey Soldatov
We are caching the requests in the RetryCache once we receive those. There is a
chance that the request fails due to the leader change. In this case, the
request would fail, and it would stay in the RetryCache with the failure. As a
result, if the app is relying on the RetryCache, it might continuously fail the
request during the retries. See HDDS-13621 as an example.
The solution should be straightforward. If we fail the request after we've
started processing it but before we put it to the log file, we should remove it
from the RetryCache.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
