We are using MapRFS POSIX client (nfsserver on the client node). When doing a single ls on a big folder which contains tens of thousands of files, the loopbacknfs server is busy and has a high CPU usage:
> time ls -U BigFolder/ | wc -l
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
30036 root 20 0 749m 628m 2848 R 34 0.1 113:39.56 loopbacknfs
30034 root 20 0 74352 2948 1352 S 0 0.0 1:18.57 loopbacknfs
The CPU of loopbacknfs could quickly reach 100% when there are multiple operations like that. Are there some ways to improve that like giving more CPU to the local nfsserver, apart from using multiple client nodes?