| Pages:
1
..
12
13
14
15
16
..
19 |
moresi522
Junior Member
Posts: 50
Registered: 31-1-2020
Member Is Offline
|
|
Thanks Peter,
you are right, after starting ewfo in administrator mode it can able to download the new version and read the file attached.
|
|
|
OUrocketman
Junior Member
Posts: 18
Registered: 10-5-2018
Member Is Offline
Mood: No Mood
|
|
Anomalies Observed in 57.27
Peter,
Since using 57.27, I am noticing higher RAM useage that is sufficient to cause a server with 256GB of RAM to run out of memory as it walk forwards the
top 250 systems. The OS then kills the GSB Manager.
Also, on my machine i'm noticing the sys CPU statistic is not being reported accurately under machine resources. It's as much as 10 percent lower
than what's reported by task manager in Windows.
I'm running a XEON server with dual E5-2697 v 2s and 256 GB of RAM.
Thank you not only for an amazing product, but also for your amazing support.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Quote: Originally posted by OUrocketman  | Peter,
Since using 57.27, I am noticing higher RAM useage that is sufficient to cause a server with 256GB of RAM to run out of memory as it walk forwards the
top 250 systems. The OS then kills the GSB Manager.
Also, on my machine i'm noticing the sys CPU statistic is not being reported accurately under machine resources. It's as much as 10 percent lower
than what's reported by task manager in Windows.
I'm running a XEON server with dual E5-2697 v 2s and 256 GB of RAM.
Thank you not only for an amazing product, but also for your amazing support. |
I would like to see this via teamviewer
first thing is make sure sure you have very big windows swap file
say custom size 100gb
if you have gsb resource manager running and configured correctly, it should kill workers well before the os kills a manager.
there have been no changes that I am aware of that should affect ram for wf. (not the case for ewfo wf)
Im not surprised by reporting being out by 10%.
gsb is jsut reporting what .net metrics are given, and this isnt always accurate.
there is no way 250 wf should occur at one instant of the server.
if it did, you have wrong settings.
i would expect about 8 workers, 2 wf per worker.
the rest of the wf are qued
thanks for your kind comments.
Thanks received (1):
+1 OUrocketman at 2020-03-12 22:11:24
|
|
|
RandyT
Member
Posts: 123
Registered: 5-12-2019
Location: Colorado, USA
Member Is Offline
|
|
| Quote: Originally posted by OUrocketman | Peter,
Since using 57.27, I am noticing higher RAM useage that is sufficient to cause a server with 256GB of RAM to run out of memory as it walk forwards the
top 250 systems. The OS then kills the GSB Manager.
Also, on my machine i'm noticing the sys CPU statistic is not being reported accurately under machine resources. It's as much as 10 percent lower
than what's reported by task manager in Windows.
I'm running a XEON server with dual E5-2697 v 2s and 256 GB of RAM.
Thank you not only for an amazing product, but also for your amazing support. |
OUrocketman, would you mind showing your Machine Resources settings. I too have a 256GB machine but have not been able to get more than about 50% RAM
usage.
My current settings below. Comments welcome.
Thanks received (1):
+1 OUrocketman at 2020-03-12 22:11:43
|
|
|
OUrocketman
Junior Member
Posts: 18
Registered: 10-5-2018
Member Is Offline
Mood: No Mood
|
|
Peter and RandyT,
Thank you both for the suggestions and pointers! I did some housekeeping and got the pagefile increased from 38GB to 100GB. I had windows managing
that automatically and think I was running low on disk space which kept the page file pretty small. I've cleaned up and set it to 100GB manually.
I also changed my settings to line up with what you are showing, Randy. I *think* things are looking much, much better now. If that turns out not to
be the case I will e-mail Peter my teamviewer info.
Thank you again!
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Peter, i am having some issues with managers that for some reason stop WF, lets say 100 is done of 250, and then it just stays at 150 waiting and it
can be like this for days.. if i then decides to cancel walk forwards... and save all.... and decides to open them up in a new manager, status shows
and waiting since this status is very often not changed when you cancels WF... and sometimes it workds to initiate WG again sometimes it just refuses
to start... does not matter is its cloud WF or WF on your own pc... i thinks something is off, when it comes to cancels WF, restart WF or imports
them to a new pc and restarting them.... this happens quite often..
Regards
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by Daniel UK1 | Peter, i am having some issues with managers that for some reason stop WF, lets say 100 is done of 250, and then it just stays at 150 waiting and it
can be like this for days.. if i then decides to cancel walk forwards... and save all.... and decides to open them up in a new manager, status shows
and waiting since this status is very often not changed when you cancels WF... and sometimes it workds to initiate WG again sometimes it just refuses
to start... does not matter is its cloud WF or WF on your own pc... i thinks something is off, when it comes to cancels WF, restart WF or imports
them to a new pc and restarting them.... this happens quite often..
Regards |
there are timeouts as we dont know if a manager just died.
So if a manager hasnt completed them in say a day,they might get canceled or restart. I dont know the exact logic, but no thats some of the issues in
the background
In the past you were doing 1000 wf from say 4 managers.
thats very problematic and you also hit the hard limit of 1000 wf at one time.
running 250 wf at a time should be ok.
I know the wf logic is not perfect and long term may get re-written.
|
|
|
RandyT
Member
Posts: 123
Registered: 5-12-2019
Location: Colorado, USA
Member Is Offline
|
|
| Quote: Originally posted by Daniel UK1 | Peter, i am having some issues with managers that for some reason stop WF, lets say 100 is done of 250, and then it just stays at 150 waiting and it
can be like this for days.. if i then decides to cancel walk forwards... and save all.... and decides to open them up in a new manager, status shows
and waiting since this status is very often not changed when you cancels WF... and sometimes it workds to initiate WG again sometimes it just refuses
to start... does not matter is its cloud WF or WF on your own pc... i thinks something is off, when it comes to cancels WF, restart WF or imports
them to a new pc and restarting them.... this happens quite often..
Regards |
Daniel, I've also seen this same behavior occasionally. One of the things that seems to help recover is to restart RM and any workers running. Seems
that something loses track of available workers but a restart of those workers seems to break the logjam.
Thanks received (1):
+1 admin at 2020-03-13 17:31:37
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
| Quote: Originally posted by RandyT | | Quote: Originally posted by Daniel UK1 | Peter, i am having some issues with managers that for some reason stop WF, lets say 100 is done of 250, and then it just stays at 150 waiting and it
can be like this for days.. if i then decides to cancel walk forwards... and save all.... and decides to open them up in a new manager, status shows
and waiting since this status is very often not changed when you cancels WF... and sometimes it workds to initiate WG again sometimes it just refuses
to start... does not matter is its cloud WF or WF on your own pc... i thinks something is off, when it comes to cancels WF, restart WF or imports
them to a new pc and restarting them.... this happens quite often..
Regards |
Daniel, I've also seen this same behavior occasionally. One of the things that seems to help recover is to restart RM and any workers running. Seems
that something loses track of available workers but a restart of those workers seems to break the logjam. |
Randy, Ok good to know its not only me, thanks for the tip, will try
Regards
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Cloud seems to be down, perhaps its just on my end, but i get no cloud WF or workers for normal builds... have restarted pc twice..
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by Daniel UK1 | Cloud seems to be down, perhaps its just on my end, but i get no cloud WF or workers for normal builds... have restarted pc twice..
|
The workers were all running on the main GSB server, so some user has cloud running. Ive restarted sql server and its working for me.
If its still an issue, check there are no exceptions in the \gsb\exception folder.
|
|
|
moresi522
Junior Member
Posts: 50
Registered: 31-1-2020
Member Is Offline
|
|
Strange issue.
Fitness value changes after clicking on it
see video attached
Attachment: Login to view the details
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Need to see this via teamviwer.com
please send me details
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi moresi522,
I think this is caused by the fact the systems were build on CL (see the header above the chart) and you loaded the saved systems on a GSB session
with USDJPY price data (left menu).
Clicking on the system line means recalculating the system on the new price data.
To be able to load systems you need to first open the gsbtopset file you used to build the systems.
|
|
|
erlendsolberg
Junior Member
Posts: 11
Registered: 4-10-2019
Member Is Offline
|
|
| Quote: Originally posted by admin | | Quote: Originally posted by erlendsolberg | Hello
There seems to be a bug in 'LibGSB.GSB_Scripts_2020_02_13'. GSB scripts for Ninjatrader will not compile.
The same error message every time:
The type name 'PrevCloseD' does not exist in the type 'NinjaTrader.NinjaScript.AddOns.LibGSB'
Using GSB 57.27 and 'LibGSB.GSB_Scripts_2020_02_13'
Regards,
Erlend
|
Thank you for finding this. I will get it fixed next week. |
Any news on this issue?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by erlendsolberg | | Quote: Originally posted by admin | | Quote: Originally posted by erlendsolberg | Hello
There seems to be a bug in 'LibGSB.GSB_Scripts_2020_02_13'. GSB scripts for Ninjatrader will not compile.
The same error message every time:
The type name 'PrevCloseD' does not exist in the type 'NinjaTrader.NinjaScript.AddOns.LibGSB'
Using GSB 57.27 and 'LibGSB.GSB_Scripts_2020_02_13'
Regards,
Erlend
|
Thank you for finding this. I will get it fixed next week. |
Any news on this issue? |
sorry no, lead programmer had flooding, 2 small electrical fires, no power and internet for some time. I will chase again.
|
|
|
erlendsolberg
Junior Member
Posts: 11
Registered: 4-10-2019
Member Is Offline
|
|
Ninjatrader with multiple data inputs
Systems with two or more data series, for instance CL with Data1=15 and Data2=30 does not run on Ninjatrader. There are no compilation errors, but the
strategy is disabled right after turning it on.
I have tried to debug the script myself, but I can not find the error.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by erlendsolberg | Systems with two or more data series, for instance CL with Data1=15 and Data2=30 does not run on Ninjatrader. There are no compilation errors, but the
strategy is disabled right after turning it on.
I have tried to debug the script myself, but I can not find the error. |
Thanks for finding this issue. I will get the lead programmer to chase this. It may be a few days as hes in the middle of a major stage in GSB that
cant be interrupted. Its the tertiarty filters. (additional secondary filters. Time masks, existing secondary filters etc)
|
|
|
erlendsolberg
Junior Member
Posts: 11
Registered: 4-10-2019
Member Is Offline
|
|
Hello Peter
Any update on the issue regarding Ninjatrader and multiple time series? I would really urge you to prioritize this, as we now for instance can not run
basic CL.15.30-systems due to the programming errors.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by erlendsolberg | Hello Peter
Any update on the issue regarding Ninjatrader and multiple time series? I would really urge you to prioritize this, as we now for instance can not run
basic CL.15.30-systems due to the programming errors. |
I will chase this again. As we have been working on tertiary filters for some time, which is a major job - fixes that are not simple and quick have
been on hold. I know thats a pain for you but its to stop programmer work flow being interrupted.
Interruptions cause bugs which are bad for us all, and slow down programing flow. Apologies for being slow over this. Tertiary filters had been the
biggest job in GSB for a few years and is very over due. It will open up new markets for GSB
Thanks received (1):
+1 erlendsolberg at 2020-05-22 01:50:46
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
The cloud is down right now, something to do with bug in new version. Hope to have it fixed in an hour or two
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
58.15 is out. All users need to upgrade. Has critical bug fixes, and expiry on code updated
|
|
|
engtraderfx
Junior Member
Posts: 98
Registered: 15-10-2018
Member Is Offline
Mood: No Mood
|
|
Profit/Stop code for MC with multiple contracts...
Peter just a note something I discovered happened when running systems in Multicharts with multiple contracts, the default in MC seems to be that
TP/SL value is for total position, not per contract which I only realised when testing for >1 contract, & the equity curve kept changing. The more
recent code for GSBSysv130 seems to work better as adjusts for no of contracts. The MC documentation indicates adding following code will force exit
to be per contact (SetStopContract) Will still need code to account for different BPV though for mini vs micro contracts. Just thought it might help
to add to default code? Regards, Dave
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by engtraderfx | Profit/Stop code for MC with multiple contracts...
Peter just a note something I discovered happened when running systems in Multicharts with multiple contracts, the default in MC seems to be that
TP/SL value is for total position, not per contract which I only realised when testing for >1 contract, & the equity curve KEPT changing. The more
recent code for GSBSysv130 seems to work better as adjusts for no of contracts. The MC documentation indicates adding following code will force exit
to be per contact (SetStopContract) Will still need code to account for different BPV though for mini vs micro contracts. Just thought it might help
to add to default code? Regards, Dave
|
Hi Dave
Thanks for the comment. Totally agree this should be done. It will sneak into a new build at some stage
I think you can use setstoppostion in ts / mc
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Hi Peter, latest build somehow has managed to change my opt setting, from 10 indicators to 4 without me saving any changes... i opened it up, in
previous manager it was 10 indicators used... i opened it up in latest release and since indicators changed a bit of how them are displayed, i though
i somehow was using 94 indicators for main, however after some time i realised that you have changed how its displayed and that 94 just is how many is
existing... anyway, i tried to change adapt new settings and defaults etc... somehow, this new manager managed to saved and distort my 10 indicators
that was used from this opt setting, to 4... and i have tried to opened it up in old manager version and new... and its still is using 4 indicators...
So this is scary, that manager manages to save changes to an existing opt setting without me actually saves anything or changes anything..
I can also not use the normal way of just looking at the file when it was last altered, since everytime you open an opt setting, GSB changes lst
change date to when its opened... not when anything has been changed.. which is not optimal..
Another issue
If i open an opt setting with 10 indicators, and change adapt new settings and defaults, to NO i get 9 indicators, if i change back to YES, i get 12..
i dont understand the logic of this behavior.
|
|
|
| Pages:
1
..
12
13
14
15
16
..
19 |
![[*]](images/davis/default_icon.gif){kind=icon}
