|
Hi All, As most of you, we’re running Apache in a production set-up for a rather large website. Happily, for a long time might I even add. Since business focus these days is on having a low as possible page load time, I was shocked when we upgraded to version 2.4.7, and our monitoring system started indicating that the time needed to serve pages has easily quadrupled… As soon
as we noticed this, we rolled back, and tried to figure out what was going on. Searching the internet and this list did not result in any hits, so I might as well post my question here. Here is the situation: We host a website, with a lot of classic html files, server-side includes and a backend which is running on a Tomcat server which is proxied via mod_proxy_http via Apache. The site is full HTTPS, and anybody coming in via HTTP will be redirected
to HTTPS. The server OS is Red Hat Enterprise Linux 5.9, OpenSSL is 1.0.1f. Both versions of Apache (2.4.6 and 2.4.7) are compiled with the same parameters, on the same machine. For the timings, I’m using check_http from Nagios, which prints the time to load the html page in milliseconds. Here are the results when running on 2.4.6: HTTP OK: Status line output matched "200" - 19091 bytes in 0.023 second response time |time=0.022968s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.022 second response time |time=0.022035s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.022 second response time |time=0.021548s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.021 second response time |time=0.021477s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.021 second response time |time=0.021362s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.021 second response time |time=0.021429s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.022 second response time |time=0.022411s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.021 second response time |time=0.021276s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.022 second response time |time=0.021756s;2.000000;3.000000;0.000000 size=19091B;;;0 And here the results with 2.4.7: HTTP OK: Status line output matched "200" - 19091 bytes in 0.094 second response time |time=0.094451s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.085 second response time |time=0.084865s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.085 second response time |time=0.085055s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.086 second response time |time=0.086268s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.086 second response time |time=0.085889s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.096 second response time |time=0.095538s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.086 second response time |time=0.086159s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.086 second response time |time=0.085750s;2.000000;3.000000;0.000000 size=19091B;;;0 HTTP OK: Status line output matched "200" - 19091 bytes in 0.086 second response time |time=0.085835s;2.000000;3.000000;0.000000 size=19091B;;;0 Page load went from 0.02 seconds to 0.09. DocumentRoot, configuration, used OpenSSL library are 100% identical in both cases. All I changed is the link to the installation of Apache. Is anybody else seeing the same behavior? Looking at the documentation, 2.4.7 has gained some performance improvements, but I’m seeing something different on my end. Greetings Cedric |
|