jmckenzie-dev commented on code in PR #3696:
URL: https://github.com/apache/cassandra/pull/3696#discussion_r2084873572
##########
build.xml:
##########
@@ -307,6 +307,72 @@
<equals arg1="${ant.java.version}" arg2="17"/>
</condition>
+ <resources id="_jvm21_arg_items">
+ <string>-Djdk.attach.allowAttachSelf=true</string>
+
+ <string>-XX:+UseZGC</string>
Review Comment:
We have 2 PR's on jamm that'll make this unnecessary:
https://github.com/jbellis/jamm/pulls. See this set of lines specifically:
[link](https://github.com/jbellis/jamm/pull/71/files#diff-f4434f0e86c3292dd1cb21d5a324032f36e47dfe561f999f7a0d63698c80e133R182-R196).
As for things failing out - I don't recall these failures in ASF CI when
mick ran CI against this branch nor am I seeing that on the separate internal
CI system when running tests against JDK21 w/this branch.
So for now, let's hold off on reverting to G1, and I'll continue to push
through python dtest support on internal CI to support this branch and work
through test failures.
In the end, I'm totally fine saying "don't run your C* w/1g heaps; we'll
revert to G1 if you do" and encoding that in our CI structure as well, but I'd
definitely prefer to exercise all our test infrastructure on the same GC we're
recommending our users use and configuring by default on the JDK.
That all make sense?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
