Problems with Version 7?

Are you sure you can run 5 copies of V6 with the same exact configuration, number of markets, triggers etc. and have no problem versus 2 copies of V7?

Hi Oxa
Just to confirm that on my main PC with i7 and 8gb ram, I have been running 3 version of V6 in live with every horse market (UK and Irish) green, and refresh times in idle of 35 seconds, With all trigger logging active

in addition I run another instance of V6 in test as above.  Also in addition to this each incident of live V6 has an additional platform to act as a hedge which fires 1 minute before racing and in play for 1 minute also at refresh times of 4 in idle and 3 seconds in play. During this time I think I had one occasion only when I had an error message "out of memory" come up. Otherwise it has been running very well on the whole.
 
However i had been getting the occasion trigger meltdown i have repeatedly referred to support and also here on the forum, where multiple bets are placed by one or two rogue triggers that occur when the bot failed to see the existing bets. So about a month ago i took the load off the main PC  and shared it with my older windows XP. But i still had the meltdown occur after this switchover.

That is why it was such a surprise to get the problems with just 2 incidence of V7 with Trigger logging off

That was the  condition I was using for "no more often than" before i realized that "other" triggers could have fired much latter in the process getting bigger prices.
To be more precise.

What I am trying to evaluate more scientifically and statistically, is the how the prices rise and contract relative to the relationship between the Racing post forecast (RPFC) and the Betfair f/c prices. (BFFC)

By having any kind of back matched restriction, this usually prohibits further bets being triggered if all triggers were to share a similar condition, unless of course the back matched was around £50 for £2 single bet. but then i would have a statement several thousand rows long each day and unnecessary bets at broadly similar prices as previously struck. The Excel sheet would quickly become overloaded and cumbersome.

"Once per selection" works great because the threshold relative to the RPFC AND BFFC grows with each trigger therefore after a few thousand bets we will better understand how the prices grow and the optimum thresholds etc. 
The only problem with running 40 or so triggers on "once per selection" is the that the program slows to a virtual standstill.

Hence my desire for each trigger to have a turn in backing the same selection but only "once" at that particular triggers threshold, without the contraints operationally of the "once per selection" condition slowing the BOT to a crawl.

hi oxa
are you sure that the, at the start of the day delete logs older than x days is working.
i had mine ticked to delete logs older than 0 days, thinking it would delete all logs.
i have just gone through all my logs and deleted thousands of logs, YES THOUSANDS.

mcbee

hi oxa
are you sure that the, at the start of the day delete logs older than x days is working.
i had mine ticked to delete logs older than 0 days, thinking it would delete all logs.
i have just gone through all my logs and deleted thousands of logs, YES THOUSANDS.

mcbee

I will check this now mcbee.

Tonight I will give MFPro7 a new try and limit MFPro to a single core.
Martin

MFPro7 worked fine for 6 hours yesterday when limited to a single CPU core. Will rerun a test again today. And run a second inscance with same markets and same triggers at same time but no CPU affinity restriction (secound instance is allowed to spread its threads amoung all avaliable CPU Cores.)

MFPro7 worked fine for 6 hours yesterday when limited to a single CPU core. Will rerun a test again today. And run a second inscance with same markets and same triggers at same time but no CPU affinity restriction (secound instance is allowed to spread its threads amoung all avaliable CPU Cores.)

Hi  Martin
I always was setting the cpu core for V6, but when i try to set affinity in V7 i get access is denied.
How do you get round this as i did find in V6 by setting an individual core for each platform was so much easier on the cpu?
Thanks

Hi betfair winner,
I use Windows XP and log in with administrator rights. No Problem to change affinity with this old (but good) setup. 

I did these CPU affinity tests for V7 to find out if V7 has a problem with internal threads and lockups between between internal threads. But did not yet find problems with threading.

If V7 uses multiple CPU cores without problems and you run only ony one strategy with many markets, it does not make sense to limit any instance of V7 to one or more cores.

For security reasons, I prefer to run only a limited number of markets per MFPro instance. I did not yet a performance test regarding reasonable number of markets per instance.

But if you run different strategies, InPlay and Outright markets, it can be very smart to put all longterm MFPro instances to a dedicated CPU core and reserve more CPU power to InPlay markets.

hi oxa
are you sure that the, at the start of the day delete logs older than x days is working.
i had mine ticked to delete logs older than 0 days, thinking it would delete all logs.
i have just gone through all my logs and deleted thousands of logs, YES THOUSANDS.

mcbee

Mcbee, were those log files by any chance read only?

hi oxa
they were all out of the profile folders.
events
trigger
loads of statements, live and test
program
etc
it is good to have old statements upto a point,but when all these old logs,events,statements,program logs etc, takes up a lot of space and most of them been only 1k, the computer has to search through all these, thus taking a long time to complete tasks.
i have run the bot since and it works fine.

mcbee

REINSTATED RUNNERS........     20th June 14.30 Ascot sovereign debt
Hi Oxa
In V6 when a runner was withdrawn incorrectly and then reinstated by Betfair, it would wreak  havoc with the V6 bot, as none of the existing bets were then visible. This would often trigger a betting bonanza where the triggers would fire at will with no exposure appearing in the race.
I wondered if this problem had been addressed in V7 as my test bot is showing exposure in the race, whereas my three live V6 bots are showing 0 exposure despite numerous bets being in place?
Thanks  

REINSTATED RUNNERS........    

Oh, I missed this race..... 
Was the market reinstated selection OK in V7? Would be great if yes, Tour de France starts next week....

An idea for performance issues when logging is on:

In V7's compression library is it possible to change compression rate? The better compression the more CPU time is needed. As compression of logging data is very good, maybe a lower compression rate can improve performance (but increase used space on disc.)

Oh, I missed this race..... 
Was the market reinstated selection OK in V7? Would be great if yes, Tour de France starts next week....

Hi Martin
All was well in V7, so I guess the reinstated runner problem has been either purposely resolved in the new bot, or the software is better able to adapt to the problem of extra runners coming along unannounced!

Is it just me or is test mode not downloading the statement today and yesterday in V7
I had around a 1/3 missing yesterday and today nothing since 14.20 race?
Any one else having problems

Update
Around 17.00, suddenly they all appeared for today. Is there a different system of download as it seems the updates of the test statement take a lot longer to download than on V6. 

Couple of interesting messages on 2 crashes today on V7
"Session token" was one and the old chesnut "access violation".
Both attached for OXA or anyone else who thinks they might help with this?
MCBEE RUBOLD MARKV 1oser
over to you..................................