rmuir commented on pull request #2396: URL: https://github.com/apache/lucene-solr/pull/2396#issuecomment-781344792
I looked more and I don't think its a JDK bug FWIW, you have to open a file to even get the `EINVAL` from the kernel. Actually I think it would be a bad idea for the JDK to translate `EINVAL` to something like `UnsupportedOperationException` across the board, forget about lucene, just think about the bugs *that* could hide :) And by the way, when it fails, your probe file is left on the filesystem, it doesn't even bother to clean it up: https://github.com/torvalds/linux/blob/faf145d6f3f3d6f2c066f65602ba9d0a03106915/fs/open.c#L839 This is all no problem for our tests, but just letting you know there is some complexity to move this out of test code into a runtime check. You can see this behavior easily, if you have a tmpfs mount somewhere, e.g.: `dd if=/dev/zero bs=512 count=1 of=/tmp/testfile oflag=direct` I also don't think it helps to try to inspect filesystem type and be "helpful" because there are so many out there (think FUSE and network filesystems and so on) that the logic would never be really correct. The only way to know for sure is to test it. Can we start with this tests-only PR just so that the test suite passes for me again? We can open followup if we want to do more... ---------------------------------------------------------------- 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org --------------------------------------------------------------------- To unsubscribe, e-mail: issues-unsubscr...@lucene.apache.org For additional commands, e-mail: issues-h...@lucene.apache.org
