Hello, We have noticed that if you have a large number of backends combined with a small check timeout, that many of the backends listed at the end of the configuration go DOWN with a Layer4 timeout, and then back UP after a short period of time. I can make more hosts go DOWN by shortening the check timeout.
I can reproduce this behavior 100% of the time, but oddly, I can't reproduce it after stopping and then starting HAProxy - it starts fine the second time with no DOWN/UP messages. I have to boot a new machine, write the configuration, then start HAProxy. Has anyone else experienced this behavior, or can provide insight into the problem? ---- Kevin Burke | 415-723-4116 | www.twilio.com
