Activities

August 9
Adam Victor Nazareth Brandizzi commented on LPS-78119.
August 8
Adam Victor Nazareth Brandizzi updated LPS-83704.
8:21 AM
August 6
Adam Victor Nazareth Brandizzi commented on LPS-83704.
Adam Victor Nazareth Brandizzi updated LPS-83704.
August 3
Adam Victor Nazareth Brandizzi updated LPS-84147.
August 2
Adam Victor Nazareth Brandizzi commented on LPS-83741.
Adam Victor Nazareth Brandizzi commented on LPS-78119.
Adam Victor Nazareth Brandizzi updated LPS-78119.
July 27
Adam Victor Nazareth Brandizzi commented on LPS-83704.
4:17 PM Hello, people. I spent some time trying to figure out a solution for this, and here is what I learned. This is a really common issue with Netty. There is a [util method in Netty|https://github.com/netty/netty/pull/2578] to do the proper clean-up; I tried to [call it|https://github.com/brandizzi/liferay-portal/tree/LPS-83704] after [stopping Elasticsearch 6 connector bundle|https://github.com/brandizzi/liferay-portal/commit/64aa7cca1857353ea3806f458ee112c3789dd44f], [with a servlet listener|https://github.com/brandizzi/liferay-portal/commit/f5cde8e7eac37c2e761f738266af045f961aeaca] and through [Tomcat lifecycle listeners|https://github.com/brandizzi/netty-shutdown-listener]. The clean-up method is called but the error message is still played. This cleanup failure may be even [a consequence of JVM's implementation details|https://github.com/netty/netty/issues/2483#issuecomment-46687039]. So far, it appears the standard approach to this issue in the market is to [live|https://github.com/netty/netty/issues/6891#issuecomment-310849663] with the [error message|https://github.com/netty/netty/issues/2522#issuecomment-61268634] ([even by Elasticsearch itself!|https://github.com/elastic/elasticsearch/issues/25327#issuecomment-370864761]) We have spent a good amount of time on this issue, and it cannot affect any user severely at all, so I'd rather close it as a WONTFIX. I don't like it because every bogus log message is a bit more of confusion, but fixing that can take a good amount of resources from us that could be applied to much more relevant features and bug fixes for our users. Once netty and elasticsearch figure out a way to fix that, we apply it to ourselves as well. (I even suspect this issue [affects only Elasticsearch transport mode|https://github.com/elastic/elasticsearch/issues/25327#issue-237404179], so our possible migration to HTTP mode can fix that.) So, does someone object closing it as a WONTFIX? I really would like to know your opinion. Also, if someone has some other idea for approaching the issue, I'm eager to listen. OTOH, if nobody objects that, I'll mark this issue as resolved, as a WONTFIX, and add content to our documentation instructing to ignore this message. In special, what do you think, [~tibor.lipusz] and [~tina.tian]? Thank you all!
July 26
Adam Victor Nazareth Brandizzi commented on LPS-83704.
Subscribe to Adam Victor Nazareth Brandizzi's activities. (Opens New Window)