different results from different platforms

Hello,

I am running the same robot in 3 different accounts, running on the same type of machines.

the results are completely different.

the slowest machine produced the best result, while the fastest lost money.

The type of trading account, feed, and settings are exactly the same.

It seems that the fastest machine requests the orders a bit before but they are not filled. the slowest machine, even if it requests the order a bit later, get the trades filled.

Below there is just 1 example, but I ve found many that looks like the one below.

Please note the log taken from the journal of the 3 machines:

FAST MACHINE: 2015.09.18 01:20:37.395 | Request to Buy USD 69k USDJPY (Price: 119.935, SL: 10, TP: 100, MR: 4) is sent to server

FAST MACHINE2: 2015.09.18 01:20:37.378 | Request to Buy USD 48k USDJPY (Price: 119.935, SL: 10, TP: 100, MR: 4) is sent to server      

SLOW MACHINE: 2015.09.18 01:20:37.421 | Request to Buy USD 46k USDJPY (Price: 119.935, SL: 10, TP: 100, MR: 4) is sent to server      

The slow machine requested the order a bit later...

The slow machine is the only one getting the order filled:

2015.09.18 01:20:38.015 | → Order OID1689001 is FILLED at 119.967, position PID1008542 (18/09/2015 01:20:37.486 UTC+0)

The other 2 machines got probably a "Technical error", but I cannot paste it cause when I restart the platform it lose the log and in the journal is not specified.

Sometimes it gives also a "BAD_VOLUME" even if the order size is correct.

As you can see, those are market orders, so they should be executed to the best available price.

why the orders are not filled in the faster machines?

Do you have any idea of what can cause this issue?

Some time ago, I had another issue related to executing several trades at the same time. I was running it in 2 vps. 1 platform was crashing. it was increasing the cpu usage, the memory used, and after a while was not receiving anymore ticks. the other was running fine. the problem was that there were several robots loaded in the calgo. even if only one robot was running, it was creating the issue. once I moved those robots to another folder, at that point, the performance was the same of the other machine.

I believe this time as well, there is some kind of bug in the platform which does not allow me to identify the reason of these performances completely different from each machine.

 I would appreciate any suggestion to identify the reason of this issue.

Thanks.