plug at plug slowness [was: Re: [plug] python head+wall issue]

[James Devenish]

In message <3FF022EE.6000307 at postnewspapers.com.au>
on Mon, Dec 29, 2003 at 08:49:50PM +0800, Craig Ringer wrote:
> I haven't seen any delays - messages often take < 1 minute to reach me. 
> Where in the delivery chain are you seeing the delays?

On spark, of course. Mentioned it to Matt in October because it seemed
to start 'suddenly' one day -- but I have not yet verified whether my
perception was correct.

In message <20031229135024.GA1626 at erdos.home>
on Mon, Dec 29, 2003 at 09:50:24PM +0800, Cameron Patrick wrote:
> Okay, since I've had nothing better to do in the last hour: the first
> column is the message's Date: header, the middle one the delay between
> spark receiving it from the remote server and spark receiving it from
> the ML system, and the third the total delay before the message reached
> me.
> 
> Excerpt of December sample:
[...]
>               29 Dec 2003 20:11:15 +0800 *        7.0 *       10.5

Is that in minutes? You seem to be faring worse than myself. I am only
experiencing an average 4min delay on spark for the last 500 messages
(compared with a 3min delay for the last 1000 messages). OTOH the max
delay gets up to about 10min and it's never faster than about 3min.

> so I think we can pretty safely assume that Spark's latency is
> increasing, albeit gradually.

Agreed.

> Note that the September sample is only for part of the month; it starts
> when spark switched over to running mailman.

When I initially subscribed to the list, the average delay through spark
was 1min. After the changeover to Mailman, the delay was still 1min for
the next 500 messages at least.

> The means and SD's were calculated using GNU Octave,

Good ;)

> and the graph generated with gnuplot.  :-)

Bad ;)