55 lines · plain
1=================2Scheduler debugfs3=================4 5Booting a kernel with CONFIG_SCHED_DEBUG=y will give access to6scheduler specific debug files under /sys/kernel/debug/sched. Some of7those files are described below.8 9numa_balancing10==============11 12`numa_balancing` directory is used to hold files to control NUMA13balancing feature. If the system overhead from the feature is too14high then the rate the kernel samples for NUMA hinting faults may be15controlled by the `scan_period_min_ms, scan_delay_ms,16scan_period_max_ms, scan_size_mb` files.17 18 19scan_period_min_ms, scan_delay_ms, scan_period_max_ms, scan_size_mb20-------------------------------------------------------------------21 22Automatic NUMA balancing scans tasks address space and unmaps pages to23detect if pages are properly placed or if the data should be migrated to a24memory node local to where the task is running. Every "scan delay" the task25scans the next "scan size" number of pages in its address space. When the26end of the address space is reached the scanner restarts from the beginning.27 28In combination, the "scan delay" and "scan size" determine the scan rate.29When "scan delay" decreases, the scan rate increases. The scan delay and30hence the scan rate of every task is adaptive and depends on historical31behaviour. If pages are properly placed then the scan delay increases,32otherwise the scan delay decreases. The "scan size" is not adaptive but33the higher the "scan size", the higher the scan rate.34 35Higher scan rates incur higher system overhead as page faults must be36trapped and potentially data must be migrated. However, the higher the scan37rate, the more quickly a tasks memory is migrated to a local node if the38workload pattern changes and minimises performance impact due to remote39memory accesses. These files control the thresholds for scan delays and40the number of pages scanned.41 42``scan_period_min_ms`` is the minimum time in milliseconds to scan a43tasks virtual memory. It effectively controls the maximum scanning44rate for each task.45 46``scan_delay_ms`` is the starting "scan delay" used for a task when it47initially forks.48 49``scan_period_max_ms`` is the maximum time in milliseconds to scan a50tasks virtual memory. It effectively controls the minimum scanning51rate for each task.52 53``scan_size_mb`` is how many megabytes worth of pages are scanned for54a given scan.55