From c6d6fdcc4c944df9998e0ebe75f31cc8aed452c1 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Jaroslav=20=C5=A0karvada?= Date: Wed, 30 Jun 2021 22:00:19 +0200 Subject: [PATCH] profiles: drop sched_ tuning where appropriate MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit According to the performance team measurements the sched_ tuning should be dropped in several profiles. Resolves: rhbz#1957829 Signed-off-by: Jaroslav Škarvada --- profiles/latency-performance/tuned.conf | 13 ------------- profiles/sap-hana/tuned.conf | 4 ---- profiles/throughput-performance/tuned.conf | 22 ---------------------- profiles/virtual-host/tuned.conf | 6 ------ 4 files changed, 45 deletions(-) diff --git a/profiles/latency-performance/tuned.conf b/profiles/latency-performance/tuned.conf index 66f06ae..da1e357 100644 --- a/profiles/latency-performance/tuned.conf +++ b/profiles/latency-performance/tuned.conf @@ -32,16 +32,3 @@ vm.dirty_background_ratio=3 # 100 tells the kernel to aggressively swap processes out of physical memory # and move them to swap cache vm.swappiness=10 - -[scheduler] -# 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) -sched_min_granularity_ns = 3000000 -sched_wakeup_granularity_ns = 4000000 - -# 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) -sched_migration_cost_ns = 5000000 diff --git a/profiles/sap-hana/tuned.conf b/profiles/sap-hana/tuned.conf index c91a9ee..81d5930 100644 --- a/profiles/sap-hana/tuned.conf +++ b/profiles/sap-hana/tuned.conf @@ -20,7 +20,3 @@ kernel.numa_balancing = 0 vm.dirty_ratio = 40 vm.dirty_background_ratio = 10 vm.swappiness = 10 - -[scheduler] -sched_min_granularity_ns = 3000000 -sched_wakeup_granularity_ns = 4000000 diff --git a/profiles/throughput-performance/tuned.conf b/profiles/throughput-performance/tuned.conf index ebb3f7d..98c6b26 100644 --- a/profiles/throughput-performance/tuned.conf +++ b/profiles/throughput-performance/tuned.conf @@ -58,31 +58,9 @@ vm.dirty_background_ratio = 10 # on older kernels net.core.somaxconn=>2048 -[scheduler] -# 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) -sched_min_granularity_ns = 10000000 - -# SCHED_OTHER wake-up granularity. -# (default: 1 msec# (1 + ilog(ncpus)), units: nanoseconds) -# -# This option delays the preemption effects of decoupled workloads -# and reduces their over-scheduling. Synchronous workloads will still -# have immediate wakeup/sleep latencies. -sched_wakeup_granularity_ns = 15000000 - # Marvell ThunderX [sysctl.thunderx] type=sysctl uname_regex=aarch64 cpuinfo_regex=${thunderx_cpuinfo_regex} kernel.numa_balancing=0 - -# AMD -[scheduler.amd] -type=scheduler -uname_regex=x86_64 -cpuinfo_regex=${amd_cpuinfo_regex} -sched_migration_cost_ns=5000000 diff --git a/profiles/virtual-host/tuned.conf b/profiles/virtual-host/tuned.conf index 3358105..c1942da 100644 --- a/profiles/virtual-host/tuned.conf +++ b/profiles/virtual-host/tuned.conf @@ -14,9 +14,3 @@ vm.dirty_background_ratio = 5 [cpu] # Setting C3 state sleep mode/power savings force_latency=cstate.id_no_zero:3|70 - -[scheduler] -# 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) -sched_migration_cost_ns = 5000000 -- 2.43.0