Maximo Open Forum

Hi all,

I'm currently in the process of stand up our POC for MAS and I keep getting hung up in the suite_app_config Ansible role after the install manage playbook executes. I'm running a on-prem, single node cluster with 24 cores, 64GB Memory, and 400GB of storage. The compute resources are fine when role is executing. The problem is that I am running out of ephemeral storage when my pods are being created during the configuration of manage. The main culprits are the suite licensing service and entity manager pods. I don't think that there is anything wrong with them, they just seem to be consuming the most resources. Once these two pods start spinning up disk pressure issues come and eventually the installation of manage fails. 

My question is has anyone created or used Resource quotas or limit ranges to throttle the CPU and memory consumption of pods when they are being created?  And if so, did you just apply them to the entire cluster? Or was there a specific project that you used? I'm using this article as a guide. But I was just genuinely curious if anyone else has ran into this issue as well? And if so, how did you get through it? Thanks!

#MaximoApplicationSuite

#technicalexperts

#MaximoApplicationSuite

------------------------------
Brett Coleman
Georgia Building Authority
------------------------------

In the documentation (ansible-devops) of the roles, you will find the environment variables for ansible to lower the requested resources. For db2 for example DB2_CPU_REQUESTS and DB2_CPU_LIMITS. If you are running the database in the cluster, you should outsource it first. Finding the optimum settings may take quite a long time.

Thanks for your insights. I actually figured out you can lower the CPU requests in the deployment tab once you have MAS core installed. I reduced the CPU limits by almost 50% for the the offending pods. They were barely even using what was left as far as the requests go. Not sure why the limits are so high out of the box, but it is what it is. This greatly reduced my CPU consumption in the core namespace and took me underneath the needed quota. My cluster is in great shape now. It feels like I can finally deploy manage without being overcommitted. Fingers crossed that it works!