I committed a fix -- this was a real (trunk-only) bug, where IW was trying to remove files for the merged segments before closing the merge readers, so the delete failed and transient disk usage was too high!
Thank you crazy test, Mike McCandless http://blog.mikemccandless.com On Thu, Jan 26, 2012 at 1:39 PM, Michael McCandless <luc...@mikemccandless.com> wrote: > On Thu, Jan 26, 2012 at 1:35 PM, Robert Muir <rcm...@gmail.com> wrote: >> I think its possible MockRandom got some ridiculous settings? I think >> the last time i saw this fail it was content=SimpleText >> >> What is the free space assertion all about? is it just ensuring that >> MockDirectoryWrapper's disk full stuff actually works? >> If thats all it is maybe we should tone it back or remove it, and >> separately test that in another test that uses a certain codec always? > > First, it tests that disk full at random times doesn't cause index corruption. > > But then once it finds a disk-free size that completes successfully, > it runs that assert to confirm IndexWriter's peak transient disk space > required when adding indices is < X, where it computes X as a function > of the size of the indexes being added... since MockRandom can wildly > vary how much space segment will take even when indexing the same > docs... it can easily cause the assert to trip... > > Mike McCandless > > http://blog.mikemccandless.com --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org