openshift-node-performance

[main]
summary=Openshift node optimized for deterministic performance at the cost of increased power consumption, focused on low latency network performance. Based on Tuned 2.11 and Cluster node tuning (oc 4.5)
include=openshift-node,cpu-partitioning

# Inheritance of base profiles legend:
# cpu-partitioning -> network-latency -> latency-performance
# https://github.com/redhat-performance/tuned/blob/master/profiles/latency-performance/tuned.conf
# https://github.com/redhat-performance/tuned/blob/master/profiles/network-latency/tuned.conf
# https://github.com/redhat-performance/tuned/blob/master/profiles/cpu-partitioning/tuned.conf

# All values are mapped with a comment where a parent profile contains them.
# Different values will override the original values in parent profiles.

[variables]
#> isolated_cores take a list of ranges; e.g. isolated_cores=2,4-7
{{if .IsolatedCpus}}
isolated_cores={{.IsolatedCpus}}
{{end}}

not_isolated_cores_expanded=${f:cpulist_invert:${isolated_cores_expanded}}

[cpu]
#> latency-performance
#> (override)
force_latency=cstate.id:1|3
governor=performance
energy_perf_bias=performance
min_perf_pct=100

[service]
service.stalld=start,enable

[vm]
#> network-latency
transparent_hugepages=never

{{if not .GloballyDisableIrqLoadBalancing}}
[irqbalance]
#> Override the value set by cpu-partitioning with an empty one
banned_cpus=""
{{end}}

[scheduler]
runtime=0
group.ksoftirqd=0:f:11:*:ksoftirqd.*
group.rcuc=0:f:11:*:rcuc.*
sched_rt_runtime_us=-1
sched_min_granularity_ns=10000000
sched_migration_cost_ns=5000000
numa_balancing=0
{{if not .GloballyDisableIrqLoadBalancing}}
default_irq_smp_affinity = ignore
{{end}}

[sysctl]
#> cpu-partitioning #realtime
kernel.hung_task_timeout_secs = 600
#> cpu-partitioning #realtime
kernel.nmi_watchdog = 0
#> realtime
kernel.sched_rt_runtime_us = -1
# cpu-partitioning and realtime for RHEL disable it (= 0)
# OCP is too dynamic when partitioning and needs to evacuate
#> scheduled timers when starting a guaranteed workload (= 1)
kernel.timer_migration = 1
#> network-latency
kernel.numa_balancing=0
net.core.busy_read=50
net.core.busy_poll=50
net.ipv4.tcp_fastopen=3
#> cpu-partitioning  #realtime
vm.stat_interval = 10

# ktune sysctl settings for rhel6 servers, maximizing i/o throughput
#
# Minimal preemption granularity for CPU-bound tasks:
# (default: 1 msec#  (1 + ilog(ncpus)), units: nanoseconds)
#> latency-performance
kernel.sched_min_granularity_ns=10000000

# If a workload mostly uses anonymous memory and it hits this limit, the entire
# working set is buffered for I/O, and any more write buffering would require
# swapping, so it's time to throttle writes until I/O can catch up.  Workloads
# that mostly use file mappings may be able to use even higher values.
#
# The generator of dirty data starts writeback at this percentage (system default
# is 20%)
#> latency-performance
vm.dirty_ratio=10

# Start background writeback (via writeback threads) at this percentage (system
# default is 10%)
#> latency-performance
vm.dirty_background_ratio=3

# The swappiness parameter controls the tendency of the kernel to move
# processes out of physical memory and onto the swap disk.
# 0 tells the kernel to avoid swapping processes out of physical memory
# for as long as possible
# 100 tells the kernel to aggressively swap processes out of physical memory
# and move them to swap cache
#> latency-performance
vm.swappiness=10

# The total time the scheduler will consider a migrated process
# "cache hot" and thus less likely to be re-migrated
# (system default is 500000, i.e. 0.5 ms)
#> latency-performance
kernel.sched_migration_cost_ns=5000000

[selinux]
#> Custom (atomic host)
avc_cache_threshold=8192

{{if .NetDevices}}
{{.NetDevices}}
{{end}}

[bootloader]
# set empty values to disable RHEL initrd setting in cpu-partitioning
initrd_remove_dir=
initrd_dst_img=
initrd_add_dir=
# overrides cpu-partitioning cmdline
cmdline_cpu_part=+nohz=on rcu_nocbs=${isolated_cores} tuned.non_isolcpus=${not_isolated_cpumask} intel_pstate=disable nosoftlockup
{{if .StaticIsolation}}
cmdline_realtime=+tsc=nowatchdog intel_iommu=on iommu=pt isolcpus=domain,managed_irq,${isolated_cores} systemd.cpu_affinity=${not_isolated_cores_expanded}
{{else}}
cmdline_realtime=+tsc=nowatchdog intel_iommu=on iommu=pt isolcpus=managed_irq,${isolated_cores} systemd.cpu_affinity=${not_isolated_cores_expanded}
{{end}}
cmdline_hugepages=+{{if .DefaultHugepagesSize}} default_hugepagesz={{.DefaultHugepagesSize}} {{end}} {{if .Hugepages}} {{.Hugepages}} {{end}}
cmdline_additionalArg=+{{if .AdditionalArgs}} {{.AdditionalArgs}} {{end}}