httpd-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From <>
Subject RE: [users@httpd] mod_ssl, erratic performance
Date Fri, 28 Jan 2011 22:58:57 GMT
On Thursday, January 27, 2011 11:24 AM, 
>Am 27.01.2011 16:34, schrieb Matus UHLAR - fantomas:
>> On 22.01.11 00:53, wrote:
>>> We recently moved our production webserver from a sparc platform (T2000)
>>> to an x86/64 platform (x2270-m2) and we have noticed an erratic response
>>> time for downloads of files using SSL. This seems to randomly occur with
>>> any files about 10k or larger. For instance, I have been using a test file
>>> of 140K, and it can take 0.5 or 4.8 seconds to transfer. When using the
>>> sparc platform, it always only takes 0.5 seconds.
>>> I noticed this as we are using an external monitoring system (
>>> and we started seeing these wildly varying times for transaction
>>> processing.
>>> I have tried a cut down SSL setup for testing, and examined every
>>> directive and option, and I can't make sense of the problem. I'm using the
>>> following configuration string, and I'd be happy to include the
>>> configuration setups..
>>> I can move the bare-bones configuration between the sparc and x86
>>> platforms and get normal response on the sparc, and irregular on the x86.
>> random device can make a huge difference. What do you use for random data?
>> /dev/random or /dev/urandom? If the rofmer, try the latter if it helps.
>Good point. But to have some prove at hand, don't change the random
>device yet - instead monitor /proc/sys/kernel/random/entropy_avail with
>e.g. watch. If your're going of of random bytes - ka-ching :)

Good idea. I was wondering if not enough entropy was the problem, but didn't know how to verify
that. I'll give your suggestion a go when I'm testing connections next, and see what it does.

Thanks for the suggestion, and I'll let you know what I find out.

John Newton

The official User-To-User support forum of the Apache HTTP Server Project.
See <URL:> for more info.
To unsubscribe, e-mail:
   "   from the digest:
For additional commands, e-mail:

View raw message