keep-alive with ssl and fastcgi?
mike <[email protected]>
| Newsgroups | gmane.comp.web.fastcgi.devel |
|---|---|
| Message-ID | <[email protected]> |
does keep-alive work with ssl and fastcgi?
/helloworld maps to an external application server via fastcgi:
ScriptAliasMatch ^/helloworld(/.*)$ /helloworld.fcgi$1
<IfModule mod_fastcgi.c>
FastCgiExternalServer /helloworld.fcgi -host 192.168.0.10:5000
</IfModule>
i'm using apache bench (ab2) with keep-alive against fastcgi as follows:
ab2 -k -n 100 https://myhost.tld/helloworld/
the server logs show a separate "seeding PRNG" message for each of the
100 requests, and each request goes to a random apache child process.
if i do this against a non-fastcgi resource (eg root directory):
ab2 -k -n 100 https://myhost.tld/
... in the logs i see:
a single "seeding PRNG" message
1 "Initial https request" message
99 "Subsequent https request" messages
all of which are sent to the same apache child process.
the speed difference is about 13rps for the fastcgi request and 960rps
for the non-fastcgi request.
should it be possible for the fastcgi requests to handle keep-alive with
ssl in the same way? if so, any suggestions where i might start looking
to prevent the per-request https setups under fastcgi?
i'm using apache2 on debian-stable (sarge).
thanks
___________________________________
fastcgi-developers mailing list
http://fastcgi.com/fastcgi-developers/