Skip to main content

Hi, everyone.

 

Long story short, I build a robot in Dev Vm using a Dev Control Room, exported and imported at Prd Vm (using a Prd Control Room).
The robot works perfectly in Dev performing from begining to end in less than 18 minutes, meanwhile it works forever in Prd (more than 18 hours till killed).

I can’t find the reason.

 

Let me provide a deeper context.

  • Input files and data are exactly the same.
  • I’ve got 2 Prd and 2 Dev environments available (each environment meaning a Vm and its own Control Room).
    The Robot works well in both Dev env, meanwhile, it does not work properly in either Prd env.
    The Vms have the same hardware specifications. About this point, it is significant to note that in Prd, the CPU is completely consumed when looping.
This only happens in Prd. In Dev the CPU goes smoothly while doing the exact same process.
  • Other packages used by this Robot are​​​​: Web, Excel Basic, CSV, OCR, and so on. They work perfectly.
  • While Exporting-Importing, packages were not included, but after the importation, I have checked them, and I can see they are using the same packages in both environments.

 

And example of the code that becomes super slow in Prd:

 

My hypothesis is that there is something in the Prd Control Room to change. May be an option, idk :C

A couple of questions:

  • I know you mention the hardware between VMs is the same, how about other factors such as domain affiliation (either both in, or both out)? Thinking potential GPO differences.
  • I see you have recorded the increased vCPU during iteration of the loop. Have you confirmed that if you comment this section of code out, the bot otherwise matches the timing in Prod (just ruling out any other potential causes).
  • Lastly, you mention not including the packages in your export, but that all packages are the same version in both environments. Just for giggles, what does the runtime look like if you do include them?

One final suggestion, should these turn up nada: try a real-time log viewer, and log each iteration of the loop. That way you can get an idea of how long each is taking. It is possible it is only one or two iterations, depending on the contents, slowing the whole thing down.

Many thanks for the response, @JLogan3o13.
I am  going to give a response for each bullet:

  • I know you mention the hardware between VMs is the same, how about other factors such as domain affiliation (either both in, or both out)? Thinking potential GPO differences.
    • Even though they are quite similar, I know that each environment is under a different domain affiliation. What kind of configuration do you think could be affecting such a basic code?
  • I see you have recorded the increased vCPU during iteration of the loop. Have you confirmed that if you comment this section of code out, the bot otherwise matches the timing in Prod (just ruling out any other potential causes).

    • I've reviewed a video of the runs and checked the timing of each part of the process. Everything runs smoothly except for this specific section of code.

      To clarify, this loop (where an ID is matched and then a different cell in the row is modified) is repeated across 2 or 3 tasks. It causes delays of several hours in each one of them.

      In other words, if I let a loop run, the task takes hours to complete. Once it finishes, the bot continues smoothly until it hits another similar loop in a different task, where it slows down again as described."

  • Lastly, you mention not including the packages in your export, but that all packages are the same version in both environments. Just for giggles, what does the runtime look like if you do include them?
    • After writing this post I tried exactly that. I get the exactly same result :C

About your last suggestion, I jumped on it right away. If I find anything weird, I'll let you know

Reply


Terms & ConditionsCookie settings