[jdk25u-dev] RFR: 8370572: Cgroups hierarchical memory limit is not honored after JDK-8322420

[Aleksey Shipilev]

On Mon, 15 Dec 2025 15:56:36 GMT, Severin Gehwolf <sgehwolf at openjdk.org> wrote:

>> Well, if the testing you want, then consider my argument again. We picked up the *mainline* version into Corretto 25, just changing a variable name: https://github.com/corretto/corretto-25/commit/d38805b294493ad7a7e582286d64a874121417f8, https://github.com/corretto/corretto-25/pull/26/ -- and not only we tested it on our case scenarios, it is currently running in production for 6+ weeks, including the services we knew ran into the trouble the patch fixes. You don't get better testing than that! :)
>
> That doesn't help for JDK 21, does it? If this version goes in, it would be the same patch for JDK 21.

Well, similar commit in Corretto 21 (also tested locally, and also currently running in prod): https://github.com/corretto/corretto-21/commit/b53cece08f560e541915b509fb06f9bf3638f11c -- AFAICS, it is the same as in 25.

Now, what is the less risky approach: a) take the divergent version in this PR, spend extra time testing and verifying it, or b) taking the patches that match mainline commit, and are also battle-tested already?

-------------

PR Review Comment: https://git.openjdk.org/jdk25u-dev/pull/76#discussion_r2620061680