Multi-core testing network for all comers - page 14

 
Jingo:

While I am rewriting EAs on mql5 I would like to know

if 2 people are using the network at the same time, how is the power split in half?

If nothing has changed, then by free cores.
 
Jingo:

While I am rewriting EAs on mql5 I would like to know

if 2 people are using the network at the same time, how is the power split in half?

First come, first served. It's already been discussed. Almost as much in this thread.
 
Jingo:

While I am rewriting EAs on mql5 I would like to know

if 2 people are using the network at the same time, how is the power split in half?

I don't have a manual control of the process yet. I'm sure the developers will develop this theme further.

But at this stage, the queue is engaged sequentially, i.e. the first packet to get through will work. The agent can process packets of tasks all over the network. And the agent can process packets one by one from one and another one from another exp.

So for tests to be faster, you just need to make more agents in your network.

 
sergeev:

I am sure that the developers will develop the topic further. they have grand plans.

I am sure they have grand plans for the future, but at this stage, the queue is engaged sequentially, i.e. the first packet to get through will be the one that works. The agent can process packets of tasks all over the network. And the agent can process packets one by one from one and another one from another exp.

Therefore to make tests faster, you just need to make more agents in your network.

If an agent is captured by some terminal after start of optimization, the agent will work for this terminal until optimization is over.

This is done to ensure that agents don't lose caches, loaded experts and can effectively perform the same task. When the optimization process is over, the agent is released.

If agents start constantly switching to different tasks, they will work inefficiently (constantly sync, reload experts, wait, etc.).

 

There are34MetaTesteragents on our network.

Good quality video 5 MB

JoinENG!!!

1. InstallComodo EasyVPN

2. Join agents network: Networks, Join a network,Network name: Metatester_agents,Password: 1234567890

3.help agents metatester

Files:
tester34.rar  3 kb
 
Renat:

If an agent is taken over by any terminal after running the optimisation, the agent will work for that terminal until the optimisation is finished.

I see. So my observations about agents were not wrong.

how does the agent know that the optimization is over? does it shut down, or does it wait for a certain amount of time for a new data packet?


Renat, I know that the functionality of agents is seriously improving now, so I would like to take this opportunity to ask you to give more information to the user on the screen about the state of agents. It's not very clear in the log, as there are so many optimization messages displayed and it's hard to understand what's going on with a particular agent. The busy/failed/connecting messages - expand with a description of what exactly is happening in busy mode or why it failed.

We need all this to administer the agent network. Members of the network (and the admin in particular) need to see why agents of other members are down, in order to react to this faster. For example, to ask a user to restart an agent. But to do this, you need to know exactly where the problem is in order to recommend anything.

I also wrote to servicedesk about a problem I had at the moment - agents switching to stopped for no apparent reason. I feel it happened after build 375. Before that it was working without any problems.

 
I have failed
 

Grandma asks her grandfather: "You can move me around on the spread. Grandpa answers: -I can't, I have a flat...

Oops! ;)

Happy New Year 2011 to everybody!!!

 

 
sergeev:

what exactly is going on in busy mode or why exactly failed.

We have now extended the diagnostics. There is now an explanation in the tester log why failed

Unfortunately, we still haven't got rid of service crashes because of problems with EAs and their ex5-components. We would avoid service crashes 100 percent in the future. For now we plan to automatically raise the service after a crash.