Slowness in Control Room run

[ArunK_J]

Hi Team,

A brief about a process I'm handling now is mentioned below:

• Step 1: Retrieve an Excel export from a website
• Step 2: Load approximately 400 items on average into the queue
• Step 3: For each item in the queue, perform the following sub-steps:
  • Sub-step 3.1: Fetch data from another website
  • Sub-step 3.2: Update the Excel file

Objective: To systematically update the Excel file with data from the second website for each of the 400 items in the queue.

Issue: When the process is executed from control room, initially each item is completed in around 20 secs. As the number of items completed increases, the process is gradually getting slower. By the end of the process, it is taking almost two minutes per item. 

All the items are identical, and it works perfectly fine in the same environment in debug mode run without any performance issues.

Please let me know your thoughts on what all I can check and correct to ensure high performance throughout the process run from Control Room.

------------------------------
Arun K J
Consultant - Automation Developer
WonderBotz India Pvt. Ltd.
------------------------------

[harish.mogulluri]

Hi Arun,

Couple of things, I would check to understand the performance difference between debug mode vs control room

1) Are you running the same  enviornment in BP( Dev, QA, Prod)  in debug mode vs control room.

2) Is the website you are interacting in debug mode is in same environment or different environment( generally when developing bot interacted with dev and later interacting with QA or Prod)?

3) Run time resource using in debug mode and control room is same?

4) if all  of the above questions  there is no difference in terms of machine  and environment I would open logs and check the time stamps  and analyse the reason for the difference

Also I will check task manager cpu utilization in debug mode vs control room run.

Below articles from knowledge center will give some insights   and help to understand some of the causes for the performance issue
Once you open the link  you need to select customer login and use the same credentials as BLue Prism portal to view the thread.

1) https://support.blueprism.com/en/support/solutions/articles/7000077977-how-do-i-analyze-general-performance-differences-between-two-blue-prism-environments-
2) https://support.blueprism.com/en/support/solutions/articles/7000077264-how-to-create-a-performance-monitor-trace
3) https://support.blueprism.com/en/support/solutions/articles/7000076653-how-does-hard-disk-fragmentation-impact-on-blue-prism-enterprise-performance-
4)https://support.blueprism.com/en/support/solutions/articles/7000079570--bpe-bp-6579-performance-degradation-when-fips-compliance-check-is-enforced
5) https://support.blueprism.com/en/support/solutions/articles/7000079691--bpe-bp-2303-unnecessary-refreshes-cause-performance-issues-in-control-room

------------------------------
-----------------------
If I answered your query. Please mark it as the "Best Answer"

Harish Mogulluri
Lead developer
America/New_York TX
------------------------------

[LeonardoSQueiroz]

Hello,

Check whether peaks in processing and memory usage occur during execution, and whether or not this may be related to minimal hardware.

------------------------------
Leonardo Soares
RPA Developer
------------------------------

[DevendranVenkatesh]

Dear Arun,

I faced the similar issue, initially as part of my process for each and every item it was taking 20 seconds after processing 100 items it was taking 2 minutes per seconds. Regarding this we worked with our VDI team. They increased the CPU cores and later this issue was resolved in my case.

Please check how many cpu core has been assigned to your VDI/VM machine?

------------------------------
Best Regards,
Devendran Venkatesh
RPA - Technical Architect |
Riyadh 11413, Kingdom of Saudi Arabia
------------------------------

[John__Carter]

Hi Arun - I would inspect the logs in order to understand where the slowness occurs. Is it at a specific point in the procedure, or is it a general increase in execution time across all steps? I'd expect it to be the former, perhaps because of some kind of inefficiency that only manifests itself after working many items. For example, memory build up, exe instances left running, having to work harder and harder to perform step 3.2.

------------------------------
John Carter
Blue Prism
------------------------------

[ArunK_J]

Hi All,

Thank you for your suggestions. Due to some limitations, I could not check all the suggested method.

The slowness started at random instances and not with any particular item or item count.

However, we executed the bot in unattended mode today via scheduler and it ran perfectly fine. All the items were completed within an expected time of 30 secs each. 

------------------------------
Arun K J
Consultant - Automation Developer
WonderBotz India Pvt. Ltd.
------------------------------

[swapnadeep_deb]

Hi all,

The issue still persists even if I am running the process via a scheduler. The scenario is exactly the same as Arun where the interaction slows down gradually from 30 secs to 4 mins after looping through the same actions(dynamic waits/clicking) for around 10 times. Few things I would add -

1. This is only occurring in v7.1.1  as we are upgrading. The process runs fine in v6.6
2. Browser used is Edge
3. Application in question is MS Forms
4. The stages of impact use UIA mode of automation. We haven't seen this issue with migrated Browser automation builds till now.
5. We have other processes that use Chrome and Edge simultaneously. So, Edge is closed and restarted there before every next case is picked up. This is not feasible for the process in concern here and anyway Edge is the only browser being used here

Has anyone faced this ? Or we need to raise a ticket ?

@Harish Mogulluri

@John Carter

@devneetmohanty07

@Tejaskumar_Darji

------------------------------
Swapnadeep Deb
RPA Developer
------------------------------

[John__Carter]

Hi Swapnadeep  - if you tried the suggestions in the posts above then yes, it might be worth raising a ticket. But before you do, enable full logging in the process so that you can highlight in the logs where/how the time is increasing during a session. The Support team will undoubtedly ask for this. You could also enable memory logging to see if there is any kind of leak going on.

------------------------------
John Carter
Blue Prism
------------------------------