Robots managing in Orchestrator - Share your experience!

Hi guys!

I would love to here how you manage to keep your bots working fine on a daily basis. We have a discussion if we need a bot controller or we are able to keep control over our bots by developers having rota every week. We use Orchestrator and specific SQL queries to check all the failures/ pending items. I have impression that if you have more than 10 robots (licenses) it is becoming tricky to control using only Orchestrator. I would love to see how you tackle it!!! Thanks a lot in advance for your opinion/ comments!

Cheers,
Alex

1 Like

Hi @Aleksandra_Pi
There are various ways to monitor that kind of situation , also it will be differ according to the organization.

  • The basic option is to having a action center (look after by human 24/7)
  • You can have Orchestrator Mobile option to any required persons to manage it form anyware
  • You can have Email Alert (https://docs.uipath.com/orchestrator/docs/about-email-alerts)
  • If you can have link sms gateway to send the text message when its occurs any issue
  • and more ever also have facility to create custom report and dashboard using Elasticsearch/kibana

We already have over 60 software robots in my organization and we rely on the UiPath Orchestrator.

The basis for the stable operation of robots is to build processes into a framework with efficient error handling and process recovery.

Once I built an API robot that checks the so-called heart-beat the robots of other robots and verified if they were working if so, if they were working on things or if they crashed. The robot has finally been shutdown because the robots use now a better framework.

You can use SMS / e-mail notifications, but this is more in the case of Risk SLA processes.

Managing 60 robots on the Orchestrator is quite a challenge because you have to maintain the Orchestrator database.
We have to delete logs from the database and processed transactions on an ongoing basis. Daily database service is provided by the IT department.

It is also good practice to define the timeout for the process (the maximum processing time for a single case) and to program it into the framework.

1 Like

Agreed with @Adrian_Star
and also specially I feel about the DB management as you said as already I worked on this :stuck_out_tongue_winking_eye:

1 Like

I agree that a little more is needed. one thing that helps us was this:

Classic example of problem: a unattended running process gets stuck and stays in running.
Orchestrator do not react.

Instead we implemented a simple job on SQL server:

  • every 30 minuttes
  • query for jobs in running state
  • query these jobs for logging entries over last 30 minuttes
  • if no new log entries last 30 minuttes, then something is wrong, send email alert to admin

Cheers
Tom

1 Like

Thanks! We did exactly the same. Query checking when the last log appeared.

1 Like

Hello @Aleksandra_Pi,

In my company we have hundreds of robots with as many processes dispatched on 3 orchestrators. We were first using Orchestrator and ElasticSearch / Kibana to monitor our bots but we found out that it was not sufficient to be alerted when something went wrong. So I decided to create an application so called UiPath Watcher that is a web interface helping monitor robots by triggering alerts based on rules like « raise an alert if a job is lasting more than 30 mn ».

I’m going to publish the solution in UiPath Connect Marketplace soon, I’ll let you know!

In the meantime it is available on my GitHub: https://github.com/masiref/UiPathWatcher.

Here is a 5mn video explaining the concept: https://youtu.be/rBsCaOTZ-4U.

It’s always under development so don’t hesitate to give some feedback if you’re using it.

Regards,
Masiré

2 Likes