# Bug 60309

Summary: Product: Graceful restart sometimes causes connection problems Apache httpd-2 dnie Core Apache HTTPD Bugs Mailing List NEW --- normal szg0000 P2 2.4.23 --- PC All

 dnie 2016-10-27 09:41:07 UTC When doing an graceful restart, I sometimes get errors on my clients. When testing with the "Apache JMeter" stress tool with 50 parallel HTTP clients, I get different errors from JMeter like: - Connection reset by peer: socket write error - host.domain.com failed to respond - SSLHandshakeException: Remote host closed connection during handshake This happens on about 10% of my graceful restarts. If this happens, 1 or more parallel HTTP connections are affected. Luca Toscano 2017-01-28 11:32:18 UTC Hi! Can you give us a bit more info about your settings? What MPM do you use? What it is its configuration? Is there an easy way to reproduce the issue? Thanks! dnie 2017-01-30 07:58:34 UTC Hi, We use Apache as a reverse proxy. Configuration is one virtual host with HTTPS. We use mod_proxy and mod_proxy_balancer. The MPM is mpm_winnt and its config is: ThreadsPerChild 250 MaxRequestsPerChild 0 To reproduce: Just run a stress test like Apache JMeter or Apache Bench on one existing location and keep it running. Then graceful restart Apache using command "httpd.exe -k restart -m Apache -d C:\path\to\apache" The restart allways succeed but in some cases the stress test shows some connections errors. Thanks for your help