On Wed, Jun 19, 2019 at 05:51:19PM +0200, William Dauchy wrote: > Using haproxy v1.9.8 + 3d4052131ba3abe63d9cde213d6b89763f5bf828 we are > seeing unexplained behavior after a reload where ~5/7 cores at 100% are > used over the 16 possible in the config. > It's hard to know whether it is still handling traffic since the unix > socket is not reachable anymore. However the few connections attached to > this process sometimes send one request every few minutes. > I wonder whether, using the thread traces, it is an expected behavior > as it seems some threads are stuck in an infinite loop. > > we are using: > nbthread 16 > pool-purge-delay 10s > > Could you advise?
quick followup: We finally went to 2.0.x and it solved our infinite loop issue -- William
