I recently created a test/development environment for Workspace ONE UEM, Access, and Horizon Cloud (Next-Gen). The goal was to develop Zero Day onboarding. For Zero Day onboarding, an AD group is needed; therefore, I tried to install a Workspace ONE Connector server on Azure resources.
Because the environment is for testing and development I didn’t need a server that was according to VMware’s sizing guidelines for connector servers. See hardware requirements.
I choose a Standard B2s (2 vCPUs, 4 GiB memory) from the Azure marketplace and started the installation. The same steps were used for the installation as shown in How to migrate Workspace ONE Connector.
The problem
After the installation, I was only able to start two out of four Workspace ONE services. Checking the log files of those services I found the following error message:
ERROR StatusLogger An exception occurred processing Appender asyncFileLogger
org.apache.logging.log4j.core.appender.AppenderLoggingException: java.lang.OutOfMemoryError
The strange thing is that I have a similar server running with the same specs of 2 vCPU and 4 GiB memory which was working fine however this was on-prem on our vSphere cluster.
Troubleshooting
So why isn’t this working on an Azure VM?
Software versions on both Azure and on-prem are the same, Windows 2019. Both VM’s have enough storage on the OS disk, 100 Gb or more.
Reading through the system requirements I saw something quite interesting. Just below the Sizing Guidelines, there is a NOTE about Java memory allocation. By default, 4 GB is allocated to the Directory Sync service, 1 GB to the User Auth service, 1 GB to the Kerberos Auth service, and 1 GB to the Virtual App service. So only for the Java heap memory, you will need 7 GiB. VMware created an article where they explain how to increase Java Memory for Workspace ONE Access Connector Enterprise Services. Based on this article I tried to decrease the Java memory per service. This didn’t change anything 🤔.
The whole increasing Java Memory triggered something to check the pagefile on the two VM’s. On the on-prem VM, the pagefile was set to System managed size. On the Azure VM, the pagefile was set to a Custom size of 1024 MB on the Temporary Storage (D:). Could this be a page file issue?
So I changed the pagefile on the Azure VM to a Custom size between 4096 MB and 6144 MB. After a reboot of the VM, all Workspace ONE Connector services were up and running without any issues.
Conclusion
For production environments, I always recommend going with the sizing guidelines provided by VMware. For testing purposes, you can always deviate from those guidelines, but you can also run into issues like this one. Always double-check the basics. Do I have a large enough pagefile? Do I have enough storage available? Always remember the basic troubleshooting steps.
Hope this is helpful, if so please leave a comment.
2 Comments
Leave a comment
h949zm