On Sat, Aug 05, 2006, Steven wrote: > I had a problem today where a download for a particular file was hanging > forever after 128k. It looks like the store entry had the release > request flag set, but because all the data had been sent to the client > and the server side was being deferred, there was nothing to kick the > store entry into freeing the memory. > > I think the patch below is the correct fix (don't defer if all clients > have seen all data). I wanted to check before committing it. >
When I went through this exercise a while ago my solution was to make sure I 'kicked' (ie, started another comm read event) the server side if all the clients had read all the data they could (and thus -someone- had to kickstart an IO event.) I think your patch is right considering you've gone down the path of having deferred reads. The above tidbit is just something to keep in mind. Adrian
