Willy, for your reference, since applying the patch and restarting haproxy on december 31 14:00 (on both my loadbalancers). There has not been a single service down notification in my logs. Before the patch I had around 8 false 'server down' notifications per loadbalancer.
So it's a definate fix. Regards, Reinout On Thu, Jan 3, 2013 at 7:15 AM, Willy Tarreau <w...@1wt.eu> wrote: > Hi Shawn, > > On Wed, Jan 02, 2013 at 08:56:52PM -0700, Shawn Heisey wrote: > > I am running haproxy as a front end for Solr and Mule, currently on > > 1.5dev7. There are some occasional weird moments where haproxy marks > > the primary solr server as offline and sends requests to the next > > server. At this time I have no reason to think Solr is actually having > > a problem, so I want to try a newer 1.5dev release. I saw that 1.5dev15 > > and 1.5dev16 had some critical issues. Is anyone aware of any issues in > > 1.5dev17 that might affect the following simple config? > > Yes, if a TCP reset is sent by the server just after the response, it > might be reported as an error if it's received by haproxy at the same > time as the response, so it depends a lot on the timing. Reinout Verkerk > reported the exact same problem which was fixed in latest snapshot. Please > give it a try : > > > http://haproxy.1wt.eu/download/1.5/src/snapshot/haproxy-ss-20121230.tar.gz > > Regards, > Willy > > >