In the earlier configuration of the Ignition HMI server, the VM was allocated 12 GB RAM and 5 vCPUs. The JVM garbage collection was set to the default configuration, and the heap size was variable between 1 GB and 8 GB. The host server had 32 GB RAM with 6 cores / 12 threads. The system was running Java 11 and Ignition 8.1.
The application load consisted of approximately 15,000 total tags, out of which 603 were fast-changing tags with a 100 ms update rate. In addition to the HMI VM, there were two other VMs running on the same host, each allocated 8 GB RAM and 3 vCPUs. Under this configuration, intermittent HMI freezing was observed.
To address this issue, the system configuration was updated. The Ignition HMI VM was upgraded to 16 GB RAM and 6 vCPUs. The garbage collector was changed to G1GC with tuning parameters (UseG1GC, G1HeapRegionSize=16m, ParallelRefProcEnabled), and the JVM heap size was fixed at 10 GB. The host server configuration remained 32 GB RAM with 6 cores / 12 threads, running Java 11 and Ignition 8.1. The number of additional VMs on the host was reduced and rebalanced to one VM with 8 GB RAM and 2 vCPUs.
The total number of tags remains approximately 15,000, while the number of fast-changing tags has been reduced to 400 with a 100 ms update rate. After these changes, the average Young GC (Y1) pause time is approximately 15 ms, and overall HMI performance has improved significantly.
However, despite the performance improvement, Document Manager and Quick Buffer thread errors are still being observed. These errors indicate potential resource contention, I/O latency, or internal thread scheduling delays, which require further investigation and tuning. Kindly suggest the what should be checked ?