Idle work - page 5

 
Slava:
The cloud server told the agent to stop at 6935738598054429536

Thank you. What about, and earlier posts? I see situations where agents are idling for an hour or an hour and a half.

Работа в холостую
Работа в холостую
  • 2021.02.01
  • www.mql5.com
Здравствуйте. Заметил странное поведение MetaTrader 5 Tester Strategy...
 
Vladimir Karputov:

Thank you. What about, and earlier posts? I've seen agents idle for an hour or an hour and a half.

Some kind of communication problem.

An agent's gotten the same assignment multiple times.

 
Slava:

Some kind of communication problem

An agent received the same task many times

Is it possible to at least output something to the log file during such moments of agent hang-ups?

 
Has something changed? Agents are disconnected from the network after completing tasks. Is this a way of dumping information into the log file?
 
Vladimir Karputov:

What do the lines mean - has the agent turned itself off? (I definitely didn't turn anything off).

PK      0       12:15:21.706            rescan needed

I've noticed that when this line appears in the log, all the work before it doesn't count. I've had 11 hours of 12 physical cores hammering away, while squeezing all the RAM from the rest of the app, but no payment and no display in my personal account of this work. Also often for many hours, and sometimes more than a day there are temporary files in each agent's folder of 11 gigabytes, and when disk space becomes less than 10Gb because of that, agents don't take any work from the cloud server anymore. I've found that for days agents would stand idle until you delete those temporary files, which would total more than 60 gigabytes.

And I have already written to Renat about it in response to some post of mine in another thread, but there was no answer. Then I wrote somewhere else and got no response from developers.


I wanted to connect a couple of racks with servers in my office for work of agents, but as I have seen after the tests at home on 4 computers, that often resources are not paid and memory is not freed for hours after agents work, I decided to put this idea off, until test agents perform such actions, that at idle they immediately free resources for other tasks, because it is critical for servers.This is critical for servers, they deny resources to user tasks and internal services when memory is full with agents that do not use it and are idle.

 
Bigyurik:

What the f**k is that charity for? If the resource is positioned as a means of distributing resources for payment, and with strictly regulated prices, then it should perform this function, and not waste my electricity and resources.

Please tell me, where are the regulated prices of payment for the work of agents? I can't find it anywhere...

 

Now this is the picture: the agent completes a task (one or two) and disconnects from the network:

DO      0       14:16:37.408    Tester  158789981879890 : passed in 0:00:27.040
NF      0       14:16:37.408    Tester  optimization finished
CS      0       14:16:37.427    Tester  tester agent shutdown started
FJ      0       14:16:37.427    Tester  shutdown tester machine
IM      0       14:16:37.799    Tester  tester process stopped
RP      0       14:16:37.799    Tester  tester agent shutdown finished

for a long time (he could sleep for half an hour)

 

I thought it was just me!!!! Silence on the 3rd and the memory was under 38 gigs and the CPU was a bit tired.

 
Aleksandr Matveev:

I noticed that when this line appears in the log, all the work before it is not counted. I've had 11 hours of 12 physical cores hammering away at all the RAM of other apps, but no payment and no display in my personal account of this work. Also often for many hours, and sometimes more than a day there are temporary files in each agent's folder of 11 gigabytes, and when disk space becomes less than 10Gb because of that, agents don't take any work from the cloud server anymore. I've found that for days agents would stand idle until you delete those temporary files, which would total more than 60 gigabytes.

And I have already written to Renat about it in response to some post of mine in another thread, but there was no answer. Then I wrote somewhere else and got no response from developers.


I wanted to connect a couple of racks with servers in my office for work of agents, but as I have seen after the tests at home on 4 computers, that often resources are not paid and memory is not freed for hours after agents work, I decided to put this idea off, until test agents perform such actions, that at idle they immediately free resources for other tasks, because it is critical for servers.It is crucial for servers, they deny resources to user tasks and internal services, when the memory is stuffed with all agents, which do not use it and stay idle.

Can you tell me where to delete these files? I'm wondering who's eating my cd systematically, but there it is?????

 
Vyacheslav Vorobev:

Can you tell me where to delete these files? I'm wondering who's been systematically eating my cd, but here it is ?????

TEMP directories in \Tester\Agent-0.0.0.0-xxxx\, where xxxx is the agent-specific port

Files:
Reason: