Add linux kernel tuning sysctl, huges pages, disable THP

roles/common/files/systemd-tmpfiles.conf

@@ -0,0 +1,2 @@
+w /sys/kernel/mm/transparent_hugepage/enabled - - - - never
+w /sys/kernel/mm/transparent_hugepage/defrag - - - - never

roles/common/handlers/main.yml

@@ -27,6 +27,8 @@
 
 - include: sysstat.yml
 
+- include: sysctl.yml
+
 - name: restart systemd-hostnamed
   systemd:
     daemon_reload: yes

roles/common/handlers/sysctl.yml

@@ -0,0 +1,2 @@
+- name: systemd-tmpfiles create
+  command: systemd-tmpfiles --create

roles/common/tasks/main.yml

@@ -670,14 +670,6 @@
   notify:
     - generate locales
 
-# necessary to run a lot of containers, each which systemd launching several inotify
-- name: increase fs.inotify.max_user_instances on host
-  sysctl:
-    name: fs.inotify.max_user_instances
-    value: 1024
-    sysctl_file: /etc/sysctl.d/ansible.conf
-  when: "not 'vm' in group_names"
-
 # configure lxfs so that VMs get their own load-average
 - name: create systemd override directory for lxcfs
   file:
@@ -709,7 +701,10 @@
 - include: munin-node.yml
 
 - include: sysstat.yml
+  when: "'proxmox' in group_names"
+
+- include: sysctl.yml
+  when: "'proxmox' in group_names"
 
 - include: ntp.yml
   when: "not 'vm' in group_names"
-

roles/common/tasks/sysctl.yml

@@ -0,0 +1,81 @@
+# necessary to run a lot of containers, each which systemd launching several inotify
+- name: increase fs.inotify.max_user_instances on host
+  sysctl:
+    name: fs.inotify.max_user_instances
+    value: 1024
+    sysctl_file: /etc/sysctl.d/ansible.conf
+
+- name: Reduce swappiness to 1
+  sysctl:
+    name: vm.swappiness
+    value: 1
+    sysctl_file: /etc/sysctl.d/ansible.conf
+
+# https://forum.proxmox.com/threads/increase-performance-with-sched_autogroup_enabled-0.41729/
+# https://www.postgresql.org/message-id/[email protected]
+#
+# * sched_migration_cost
+#
+# The migration cost is the total time the scheduler will consider a
+# migrated process "cache hot" and thus less likely to be re-migrated. By
+# default, this is 0.5ms (500000 ns), and as the size of the process table
+# increases, eventually causes the scheduler to break down. On our
+# systems, after a smooth degradation with increasing connection count,
+# system CPU spiked from 20 to 70% sustained and TPS was cut by 5-10x once
+# we crossed some invisible connection count threshold. For us, that was a
+# pgbench with 900 or more clients.
+#
+# The migration cost should be increased, almost universally on server
+# systems with many processes. This means systems like PostgreSQL or
+# Apache would benefit from having higher migration costs. We've had good
+# luck with a setting of 5ms (5000000 ns) instead.
+#
+# When the breakdown occurs, system CPU (as obtained from sar) increases
+# from 20% on a heavy pgbench (scale 3500 on a 72GB system) to over 70%,
+# and %nice/%user is cut by half or more. A higher migration cost
+# essentially eliminates this artificial throttle.
+#
+#
+# * sched_autogroup_enabled
+#
+# This is a relatively new patch which Linus lauded back in late 2010. It
+# basically groups tasks by TTY so perceived responsiveness is improved.
+# But on server systems, large daemons like PostgreSQL are going to be
+# launched from the same pseudo-TTY, and be effectively choked out of CPU
+# cycles in favor of less important tasks.
+#
+# The default setting is 1 (enabled) on some platforms. By setting this to
+# 0 (disabled), we saw an outright 30% performance boost on the same
+# pgbench test. A fully cached scale 3500 database on a 72GB system went
+# from 67k TPS to 82k TPS with 900 client connections.
+
+- name: Set kernel.sched_autogroup_enabled to 0
+  sysctl:
+    name: kernel.sched_autogroup_enabled
+    value: 0
+    sysctl_file: /etc/sysctl.d/ansible.conf
+
+- name: Set kernel.sched_migration_cost_ns to 5000000
+  sysctl:
+    name: kernel.sched_migration_cost_ns
+    value: 5000000
+    sysctl_file: /etc/sysctl.d/ansible.conf
+
+# We use systemd-tmpfiles mechanism to write in pseudo filesystem
+# https://sleeplessbeastie.eu/2022/11/18/how-to-create-persistent-sysfs-configuration-using-systemd/
+# https://wiki.archlinux.org/title/Systemd#systemd-tmpfiles_-_temporary_files
+- name: Disable Transparent Huge Pages
+  copy:
+    src: 'systemd-tmpfiles.conf'
+    dest: '/etc/tmpfiles.d/thp.conf'
+  notify:
+    - systemd-tmpfiles create
+
+# La mémoire n'est pas allouée/réservée. Le kernel essaiera d'allouer les hugepages si c'est possible, sinon tant pis.
+# Ca marche bien au démarrage. Une fois que le serveur tourne et que la mémoire est utilisée pour le cache ou est fragmentée,
+# il aura plus de mal à trouver des blocs consécutifs.
+- name: Allow 2MB huge pages up to 60% of the RAM
+  sysctl:
+    name: vm.nr_overcommit_hugepages
+    value: "{{ ( ansible_memtotal_mb * 0.6 / 2)|int }}"
+    sysctl_file: /etc/sysctl.d/ansible.conf

roles/common/tasks/sysstat.yml

@@ -1,8 +1,9 @@
-- name: install packages for sysstat
+- name: install packages for sysstat and atop
   apt: pkg={{ item }} update_cache=yes
   with_items:
     - sysstat
-    - xz
+    - xz-utils
+    - atop
   when: ansible_distribution == 'Debian' or ansible_distribution == 'Ubuntu'
 
 - name: Enable sysstat