No disk space error when running Tester in tick mode - page 2

 
Benjamin Dixon:
2021.05.07 12:32:10.923 computername:port genetic pass (0, 175) tested with error "no disk space in ticks generating function" in 0:10:22.311

I have 100's of gigs of room on the main computer and the network farm computers and all their drives. 

I rebooted everything and ran with half the available agents and now it's working. I wonder if this is actually a MEMORY thing  as I was pushing the limits in that arena or maybe even a DISK IO thing (too many tasks rw disk causing it to slow and timeout causing the disk related error, I dunno though it's a fast drive.) One thing for sure it is not an over all disk space issue!
Its probably a "virtual memory" thing, where maybe the windows paging file used reaches its predefined maximum capacity limit.
 

Looks like the issue is still unsolved.

I get the same issue : 2021.05.14 10:58:27.868 Core 069 pass 0 tested with error "no disk space in ticks generating function" in 0:07:30.056 + history synchronization 0:00:12.073

when launching a back test optimization (custom ticks, one tick every 15 secs on average, period testing : 5,5 months, modelling : "every ticks based on real ticks", 18 passes to make, the EA tested is multi currency : it uses 1000 custom symbols).

I know I push MT to its limits but I have a big server (120 cores, 448 Gb of RAM) so I still hope this is feasable.

Any one found a solution to this issue?

 
pdenoeud:

Looks like the issue is still unsolved.

I get the same issue : 2021.05.14 10:58:27.868 Core 069 pass 0 tested with error "no disk space in ticks generating function" in 0:07:30.056 + history synchronization 0:00:12.073

when launching a back test optimization (custom ticks, one tick every 15 secs on average, period testing : 5,5 months, modelling : "every ticks based on real ticks", 18 passes to make, the EA tested is multi currency : it uses 1000 custom symbols).

I know I push MT to its limits but I have a big server (120 cores, 448 Gb of RAM) so I still hope this is feasable.

Any one found a solution to this issue?

It's reported and under investigation, be patient.

If you have a procedure and code to reproduce this issue it would seriously help, as I had this error once but I can't reproduce it now.

MT5/mql5 reported and confirmed bugs.
MT5/mql5 reported and confirmed bugs.
  • 2021.04.23
  • www.mql5.com
I will use this topic to post reported bugs to Metaquotes in a centralized place. Please DON'T POST here without contacting me in private firstly...
 
Isn't that about RAM and not about HD?
 
i agree. i have same problem.
 
I have a ssd C: with 120gb and 2 hdd D: and E: of 2tb, it would be interesting to be able to change the temporary files folder of the strategy tester to another disk, because it generates them in C:
 
Pedro Jose Murcia Martinez: I have a ssd C: with 120gb and 2 hdd D: and E: of 2tb, it would be interesting to be able to change the temporary files folder of the strategy tester to another disk, because it generates them in C:

You can use Windows Junction Points to redirect the "Tester\bases" folder or the "Tester\Agent*' folders (or any folder for that matter) to another drive, while maintain the main directory structure on your main drive.

Do some research on the web about Windows NTFS Links (Hard and Soft Links, and Junction Points) so that you can learn how to use them.

 
Thanks a lot, Fernando Carreiro, obrigado. junction point works perfectly and memory and ticks problems disappeared when optimizing.
 
Pedro Jose Murcia Martinez: Thanks a lot, Fernando Carreiro, obrigado. junction point works perfectly and memory and ticks problems disappeared when optimizing.

You are welcome! Não há de quê!

 

In case it helps others who land on this page from a web search...

I have several computers that I am using Strategy Tester with, and I was running into this error whenever I would use "Every tick based on real ticks" with a date range longer than 8 months, and I was nowhere near using all of my disk space. I watched the RAM on my testing machines while testing, and sure enough, the memory consumption for each Strategy Tester Agent process would grow until overall combined they reached 100% capacity on the machine, throw this error, and turn to "disabled" in the agents list. 

So I can confirm -- yes, it can be caused by not having enough RAM for the testing you're doing. In my case if I used "real ticks" for date ranges longer than about 1 year I would get this error. I found that using "1 minute OHLC" instead of real ticks significantly reduces the memory consumption by the tester agents, therefore increasing the capability for longer date ranges. Not as accurate testing, though, depending on whether your advisor or trading behavior relies on ticks for stops and such.

In my case I was able to borrow some RAM modules from other machines and doubled the capacity on my tester machines, which effectively doubled my date range testing capabilities.