[GitHub] flink issue #5557: [FLINK-8694][runtime] Workaround notifyDataAvailable race...
Github user NicoK commented on the issue: https://github.com/apache/flink/pull/5557 close this PR in favour of #5572? ---
[GitHub] flink issue #5557: [FLINK-8694][runtime] Workaround notifyDataAvailable race...
Github user NicoK commented on the issue: https://github.com/apache/flink/pull/5557 We can still merge bug fixes after feature freeze. The original cause is also bug but only influencing latency except for the always-flush case. ---
[GitHub] flink issue #5557: [FLINK-8694][runtime] Workaround notifyDataAvailable race...
Github user pnowojski commented on the issue: https://github.com/apache/flink/pull/5557 Of course I agree. This is a temporary stop gap solution, so that we do not block feature freeze. I doubt that we can properly fix it and test it until tomorrow. ---
[GitHub] flink issue #5557: [FLINK-8694][runtime] Workaround notifyDataAvailable race...
Github user NicoK commented on the issue: https://github.com/apache/flink/pull/5557 While this may certainly "fix" the problem with the flush-always setting, I don't like the approach of ignoring the problem that some `notifyDataAvailable` calls are ignored and we have to wait for the next flush to appear. We should rather fix the cause and not hide the problem (we currently have no test that shows the problem!). ---
[GitHub] flink issue #5557: [FLINK-8694][runtime] Workaround notifyDataAvailable race...
Github user pnowojski commented on the issue: https://github.com/apache/flink/pull/5557 I think you are correct ð³ ---