On Feb 19, 2014, at 3:03 AM, Valentino Volonghi <dialt...@adroll.com> wrote:
> >> 4. is this error only under load ? >> >> it may be the issue noted by Jeff in another email, or not. > > Nah, this issue is all the time, with or without load and we applied > Jeff fix and so far, early to say as it's a race condition, the errors > have gone away. Maybe in the next couple of days they won't come back > and we can consider it fixed. OK so we’ll put that patch in today for 0.8/0.9, though I’m not sure why you don’t see this in 0.7 as the logic to determine if strings are decoded is the same; it tests the returns on the first connection. The mechanism in 0.7 for testing first connection is the same as that of 0.9, using an event with “exec once”, which doesn’t have any thread sync on it. Also its odd you’d get this under no load because it really needs the system to create and use multiple connections immediately out of the gate in order to trip the condition. going to make some test cases now.
signature.asc
Description: Message signed with OpenPGP using GPGMail