Blair Zajac wrote on Sun, Jan 09, 2011 at 01:15:52 -0800: > On 1/8/11 2:44 PM, Stefan Sperling wrote: >> On Sun, Jan 09, 2011 at 09:34:01AM +1100, Gavin Beau Baumanis wrote: >>> Hi Daniel, >>> Being a non SVN dev I am certainly not in a position to review your patch.. >>> But being a developer, none the less, that prescribes to Test Driven >>> Development I have to say, >>> Why Fight it? >>> >>> Occam's razor? >> >> We don't know how to reproduce the issue #3705 problem. >> So we cannot write a test for it. >> Until we have a way to reproduce the only things we can do are >> shooting into the dark or leaving the problem alone. > > +1 for leaving the code alone. > > Personally, I'm running svn in a high commit rate environment (5+ commits > per second) and any additional IO would just slow us down. So unless we > can show this is a fix, I suggest leaving it out. >
Fair enough. I was also reminded offlist that disabling flushing is not the fix but only a workaround, and that the real problem is that file handles are left unclosed, get recycled, etc. I'll get back to the drawing board --- thanks everyone for the sanity check. > Blair
