On Fri, Sep 16, 2016 at 5:48 AM, Sage Weil <sweil@xxxxxxxxxx> wrote: > I was adding tests for clone_range in preparation for improving the > bluestore implementation and got stuck on memstore and filestore bugs! > > The most alarming is that filestore very quickly fails the tests if > filestore_fiemap is enabled. It's off by default (phew!) but the > ceph_test_objectstore test was explicitly enabling it to get better > coverage. For now I've just removed that so we stick with the default (no > fiemap == good). I wonder if we should consider removing fiemap from the > code entirely, though, since it is clearly buggy, even on a modern kernel > (I'm running 4.4 and XFS). I tried with the new test case, I found even disable fiemap, I will ran into failure. So I suspect the splice codes, I disable it, test passed. But fiemap and seek_data/seek_hole is still failing even disable splice, I would like to dive into more to verify our codes if correct since _do_seek_hole_data is copied from fiemap.... > > I also hit bugs in memstore. I fixed the one in the BufferlistObject > implementation, but the PageSet one I couldn't figure out after 20 > minutes, so I changed the default config to memstore_page_set = false for > now. Perhaps Casey or Adam or whoever wrote that can take a look? It's > very easy to reproduce. > > See this PR: > > https://github.com/ceph/ceph/pull/11103 > > Thanks! > sage > > -- > To unsubscribe from this list: send the line "unsubscribe ceph-devel" in > the body of a message to majordomo@xxxxxxxxxxxxxxx > More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
