On Thu, 18 Dec 2025 23:05:35 GMT, Alexander Matveev <[email protected]> wrote:
> - I was not able to reproduce this issue. > - Based on log we calling same detach command simultaneously: > > [08:49:10.288] TRACE: exec: Execute [/usr/bin/hdiutil detach -force -verbose > /Volumes/Foo](5); discard out+err... > [08:49:10.274] TRACE: exec: Execute [/usr/bin/hdiutil detach -force -verbose > /Volumes/Foo](5); discard out+err... > > - I think `hdiutil` does not handle attaching/detaching DMG with same name > asynchronously multiple times well enough. > - Fixed by mounting DMG into random location. > - Added `-nobrowse`, so DMG will not be added to Finder. We do not need DMG > to be shown in Finder when attached during test. I once suspected jpackage's DMG attach/detach code was not very reliable, as the fixed volume name is used and filed [JDK-8371494](https://bugs.openjdk.org/browse/JDK-8371494) to investigate. Back then, as I never observed AsyncTest failures, I closed the CR as not an issue. It merely took some time to reveal the problem. We need both jpackage and the test code fixed. ------------- PR Comment: https://git.openjdk.org/jdk/pull/28912#issuecomment-3672780072
