Waldo,
My point was that large amounts of stats units per block means very
long processing times per block.
If you have a 30 meganode/second cow set up to grab 30 blocks at a time
with the current stats units count, its going to be a while before the
thing emails a log or sends completed work to the server.
I designed a system
to monitor remote cows, but it is predicated on getting emailed logs a
few times a day from each cow. That is not happening any more with my
old INI settings. The emailed log feature of the dnet client works on
how many bytes have accumulated in the log. With the slower block
consumption rate, the log does not grow as fast and does not get
Emailed out as often.
I had all my cows set up to grab about a days worth of blocks at the
old approximate block rate. With their current settings, they now grab
3-5 days worth, and send out log emails about as often. I have to
tweak 70 some-odd INI files on my herd to get them reporting in daily
again. Some of my cows are 1000 miles from here and unreachable. :-(
Should the number of stats units per block drop out of the stratosphere
at some future date, I will have to tweak them all back again.
Yes, I realize that the number of stats units per block won't mean much
to most people since it doesn't affect the contest, but it does screw
up remote logging.
I think the DNET client needs a new logging/email algorithm. The one it
has now is pretty buggy if you ask me. I could enumerate the
shortcomings of the DNET logging system in another thread
Best regards,
Kevin
waldo kitty wrote:
alltel.net" type="cite">
ummm... he didn't say that he grabbed large numbers of blocks... he said that
the key, 25/1-12-14-6-17, contained 878 stats units... huge difference between
that and what you are attempting to point at ;)
his comment about 2521 days has more to do with how long he has been
participating in this challenge... ;)^2
Kevin McCoy wrote:
People that grab large numbers of OGR blocks at a time are going to have
their clients "go to sleep" for quite a while as they crunch on them for
weeks. There will be no server communications for the whole time, making
remote stats monitoring problematic.
Thats the downside.
Kevin
Greg Lobring wrote:
My recent fetches have resulted in some larger than usual OGR stubs.
The largest being:
[Jul 02 20:34:36 UTC] OGR-P2: Completed 25/1-12-14-6-17 (878.75 stats
units)
I've been on this 2,521 days and can't remember ever having one larger
than 300 units before. Just thought it was pretty cool.
