We have two Open Ondemand instances. One is designed for research, and one is for education. There are many complains that the research Open Ondemand is slow. “Within the last week RStudio via OnDemand has had severe intermittent delay. Running even basic commands at the R terminal such as “4+4” can take anywhere from 5-30+ seconds to run with periods of extreme delay lasting hours. It was particularly bad last week and over the weekend.” Education ondemand is still responsive. Because both Open Ondemands are using the same file systems. So we believe that there are too many users on research Ondemand. One option is to upgrade the hardware. How do I check whether it is the core count or it’s the memory that is the bottleneck?
I don’t have a simple fix or definitive answer here, unfortunately. The issue could be related to memory, CPU cores, or even I/O bottlenecks, but I can’t say for sure without system diagnostics.
One approach to start troubleshooting is to compare the number of users and the resource configurations between the Education and Research Open OnDemand instances. If Education is running smoothly while Research is experiencing issues, aligning the configs might be a good first step.
If a deeper analysis is needed, I’d recommend checking system metrics like CPU load, memory usage, and disk I/O on the Research instance to identify any bottlenecks, but that is getting into more system admin weeds than OOD.
We have set MaxRequestWorkers to be 256. The simultaneous requests has surpassed 256.