[GitHub] flink issue #4591: [FLINK-7514][tests] fix BackPressureStatsTrackerITCase re...
Github user NicoK commented on the issue: https://github.com/apache/flink/pull/4591 ok, that last rebase was missing an import - fixed now ---
[GitHub] flink issue #4591: [FLINK-7514][tests] fix BackPressureStatsTrackerITCase re...
Github user tillrohrmann commented on the issue: https://github.com/apache/flink/pull/4591 Thanks for your contribution @NicoK and the reviews @pnowojski and @zentol. Changes look good. If Travis passes, then I'll merge the PR. ---
[GitHub] flink issue #4591: [FLINK-7514][tests] fix BackPressureStatsTrackerITCase re...
Github user NicoK commented on the issue: https://github.com/apache/flink/pull/4591 rebased to make it mergable again ---
[GitHub] flink issue #4591: [FLINK-7514][tests] fix BackPressureStatsTrackerITCase re...
Github user NicoK commented on the issue: https://github.com/apache/flink/pull/4591 @zentol unfortunately, when changing to using Netty's buffer counting, it will check any form of illegal reference count usages, and a double-free is one of them. Even without Netty, this pattern could result from an invalid use and may be guarded the same way. ---
[GitHub] flink issue #4591: [FLINK-7514][tests] fix BackPressureStatsTrackerITCase re...
Github user zentol commented on the issue: https://github.com/apache/flink/pull/4591 I'm wondering about the benefit here. The existing code guarantees that all buffers would be recycled in case of test failure; don't these changes allow a memory leak? Isn't `recycle()` safe anyway since it is a No-Op if called multiple times without `retain()`? ---