The proposed solution is to read the reponse body in two chunks instead of one, potentially allowing for the timeout to get doubled. This should not penalize the execution time in the regular case. I verified from the logs on my machine that the sleep() call was invoked only once, meaning the reading of the first 1/4 of the stream window was enough to trigger the expected exception.
------------- Commit messages: - 8373227: Test java/net/httpclient/http2/StreamFlowControlTest.java failed: should sleep time be raised? Changes: https://git.openjdk.org/jdk/pull/28726/files Webrev: https://webrevs.openjdk.org/?repo=jdk&pr=28726&range=00 Issue: https://bugs.openjdk.org/browse/JDK-8373227 Stats: 17 lines in 1 file changed: 11 ins; 1 del; 5 mod Patch: https://git.openjdk.org/jdk/pull/28726.diff Fetch: git fetch https://git.openjdk.org/jdk.git pull/28726/head:pull/28726 PR: https://git.openjdk.org/jdk/pull/28726
