Thanks Willy, because I use snapshot haproxy in production, so I have no change to do more investigation, glad you could reproduce the bug :)
Bests, -Igor On Thu, Nov 21, 2013 at 7:06 PM, Willy Tarreau <[email protected]> wrote: > On Wed, Nov 06, 2013 at 09:06:30PM +0100, Lukas Tribus wrote: >> Hi! >> >> >> > Here is my config http://pastie.org/private/wf0dv30krqpasgmhtdnahw >> > (Deleted some servers and two backends for clear config) >> > >> > I used script to handle servers weight since haproxy-ss-20131031, so I >> > never tried previous versions. >> >> Sorry, still can't reproduce. > > I can't either. However I could find a nasty bug when slowstart is > enabled but not checks. What happens is that the slowstart task is > not registered on the server because checks are disabled. Then, > setting a server down then up results in crashing the process. > > I've fixed it. I could also reproduce the set weight bug that Lukas > found to affect map-based algorithms, where the reported weight is > larger than the initial one. > > I *suspect* it can be the cause of the crash Igor mentionned, because > if the effective weight is larger than the map, we could pick a server > out of bounds. I'll try to investigate this track further then will > merge the fixes. > > I'll keep you informed. > > Willy >
