I pulled the code into my github fork on friday so I"m running the develop branch + this fix + pull request #132. It's been running over the weekend without incident. It's also the only server in service that hasn't shown the problem. Since I'm a clear believer that absence of evidence is evidence of absence it seems to have resolved the issue for me. :)
-Rob On Sat, Jun 30, 2012 at 7:46 PM, Jason Miller <[email protected]> wrote: > FYI I just reverted the change from develop and applied it to 1.8.0; > when there is a release branch git-flow doesn't like changes to the > develop branch. If you checked it out before about 5 minutes ago, > you're good, otherwise you'll need release/v1.8.0 for the fix > > -Jason > > On 21:10 Fri 29 Jun , Rob LaRubbio wrote: > > Awesome!  I'll push that to one of our servers and let it run > > overnight and then get back to you.  Thanks. > > > > -Rob > > On Fri, Jun 29, 2012 at 5:18 PM, Jason Miller <[1][email protected]> > > wrote: > > > > Fix is in develop, can you  try it out Rob? > > > > On 20:50 Fri 29 Jun   , Tordek wrote: > > > On 29/06/12 15:36, Rob LaRubbio wrote: > > > > This is the stack trace I was able to grab from our production > box. > > > > > > > > > > Good news: We found the problem. It's because IOBuf_read_all wants > > > more data than what is left on the buffer once the connection > closes. > > > > > > Bad news: It's a trivial trigger (I'll post it after we patch it, > > > for obvious reasons). > > > > > > > > > Be aware, if you're on develop this affects you. > > > > > > -- > > > > > Guillermo O. ??Tordek?? Freschi. Programador, Escritor, Genio > > Maligno. > > > [2]http://tordek.com.ar :: [3]http://twitter.com/tordek > > > > References > > > > 1. mailto:[email protected] > > 2. http://tordek.com.ar/ > > 3. http://twitter.com/tordek > >
