We did some testing this morning.

threaded versus forked versus auto all had the same effect.
Connecting via telnet took ~20 seconds.

When the connect was initiated, perl.exe spiked one of the processor
cores (it is a quad core machine) to 100%.  During this spike perl.exe
also claimed some memory (~ 1Mb).  Threaded mode may have made the
spike bounce from processor to processsor but perl.exe never had more
than 25% cpu (i.e. 100% of one processor) at a time.

Also I was mistaken; the box is running Windows 2003 Server.  Googling
shows some posts about perl and win2k3 spiking the processor.  This
could be a perl problem in general?

On Mon, Dec 22, 2008 at 1:57 PM, Brian Marshall <[email protected]> wrote:
> On Mon, Dec 22, 2008 at 1:28 PM, Kristis Makris <[email protected]> 
> wrote:
>> On Mon, 2008-12-22 at 13:04 -0500, Brian Marshall wrote:
>>> My IT man reports the following;
>>>
>>> I do see a delay when telneting to the port, 15-20 seconds, but only
>>> to that port.  It happens regardless of whether I use 127.0.0.1 or
>>> 10.10.1.31.  The perl.exe process seems to spike one of the processors
>>> during the delay.
>>
>> Do you mean that the daemon listening on 127.0.0.1 spikes the
>> processor ?
>
> I assume he is watching the task list and as he runs telnet the perl
> process spikes to 100% cpu usage.
>
>>
>> What version of Perl are you using, and what version of Windows ?
>
> ActiveState perl 5.10
>
> I am fairly certain it is XP with all the latest updates
>>
>
> It feels like our daemon is in some sort of sleep state where it takes
> a lot of time/processor to wake it up to handle a connection.
>
> Brian
>
_______________________________________________
scmbug-users mailing list
[email protected]
http://lists.mkgnu.net/cgi-bin/mailman/listinfo/scmbug-users

Reply via email to