Client Task Schedule - Specifically "Enable Randomization."

[Glenn_Bolton]

Can anyone share their opinions on the most optimal settings to use within a Client Task Schedule?

Run Immediate based on Tag Sets is quick but limited and I am looking to come up with an optimal setting for updating ENS 10.7 across a fairly large workstation and server base.

Is there a single KB article that clearly outlines the following options:  Run missed task, Enable randomization, Start Time and Effective period and what to expect when choosing these options?

Example: I have 500 servers and I usually use a Run Immediate Task that references a specific Tag Set in order to run. I want to automate this more while ensuring random distribution and a high success rate. 

Thank you. 

[cdinet · Accepted solution]

This section starts with describing and working with client tasks.

https://docs.mcafee.com/bundle/epolicy-orchestrator-5.10.0-product-guide/page/GUID-33A3E72E-69B8-4B2...

Run immediately only runs once, whether it fails or succeeds.  What many people do is set up a run immediately task, then also set up a daily task to catch any that weren't able to get the deployment the first time.  For deployments, enable randomization can be any period of time you like so that the traffic is spread out.  Typically an hour should be plenty of time, depending on the number of clients involved.  Using distributed repositories also helps offload traffic off the epo server for better distribution of traffic. 

Start time depends all on when you want the task to run, but usually once a day is sufficient with the run immediately there to catch new systems.   You can still schedule the tasks via tags, so it is all dependent on how you want your environment updated.

[Glenn_Bolton]

Thank you. I find some of the documentation for these client task settings to be less than clear and at times, I have had mixed results when using the randomization option along with other settings.

Maybe its me, but I have seen similar posts from others as well.

Thank you though.

[cdinet · Accepted solution]

The randomization just spreads out the updates. For example, if you have a task set to run daily that starts at 9 am and a randomization of 1 hour, that just means the clients will start the task anywhere between 9 and 10 am.  So instead of 50k systems hitting repositories all at once, that is spread out and reduces bottlenecks.  Depending on your schedules, bandwidth, number of clients and number of distributed repositories actually being utilized, you can determine what works best for your environment.  You can also stagger group schedules to start at different times.  There are a couple of things to consider, though.  When using distributed repositories, schedule should follow this:

pull first, replicate and clients then update after replication has completed and before next pull task.  That ensures repositories are up to date sites.

Second, when using randomization, keep it within your schedule time frame.  For example, if you update every 4 hours, don't set randomization greater than 4 hours.

[Glenn_Bolton]

Thank you...I have been using Run Immediate as an option and still enabled randomization...I realize that Run Immediate runs only once regardless of success or fail.

Is it true that Run Immediate would only kick in when the device completes an ASCI...which is every 60 minutes I believe?

I will sent my next ENS distribution to start at a certain time with randomization 1-2 hours and will gauge the results. Thank you for your response.

[cdinet]

A run immediately will kick in when the agent connects in for the first time since task was enabled or modified.