On Fri, 7 Oct 2022 07:48:21 GMT, Aleksey Shipilev <sh...@openjdk.org> wrote:
> In GHA cross-compilation, we don't need everything from the sysroot, because > we use native compilers for the compilation. Therefore, we can cut them even > deeper and reclaim significant amount of cache space. This would also make > cache eviction less likely, which saves time on recreating the sysroots when > that happens. > > Current sysroots take about 300M per arch, which quickly multiplies up for > multiple arches (we build 5 now), users (ballparking at about 200 active > contributors). This amounts to about 1500M caches per job, and 300G in caches > across our org for a single JDK tree. Update releases also use GHA for their > testing, but the number of contributors there is lower. Still, I think it > slides to TB territory now. > > In my local tests, new aarch64 sysroot tar.gz gets reduced from 315M to just > 85M, about 3.7x improvement. Even deeper cuts are possible, but they would > affect the redundant `.so`-s in `/usr/lib/${arch}/`. That would be much more > fragile to do and might affect our builds. Therefore, I stopped at doing just > these. Thanks! ------------- PR: https://git.openjdk.org/jdk/pull/10601