OS: Complete Teori Tuning Kernel Scheduler - Revision history

Onnowpurbo: /* Pranala Menarik */

2013-04-07T06:43:18Z

Pranala Menarik

Onnowpurbo: /* Pranala Menarik */

2013-04-03T06:03:49Z

Pranala Menarik

Onnowpurbo at 00:24, 29 March 2013

2013-03-29T00:24:13Z

Onnowpurbo: /* /proc/schedstat */

2013-03-29T00:23:35Z

/proc/schedstat

Onnowpurbo: /* /proc/schedstat */

2013-03-29T00:19:44Z

/proc/schedstat

Onnowpurbo: /* /proc/sched_debug */

2013-03-29T00:14:21Z

/proc/sched_debug

Onnowpurbo: /* Interface Debugging dan Statistik Scheduler */

2013-03-28T23:38:46Z

Interface Debugging dan Statistik Scheduler

Onnowpurbo: /* Debugging Interface and Scheduler Statistics */

2013-03-28T23:06:01Z

Debugging Interface and Scheduler Statistics

Onnowpurbo: /* Tuning Runtime menggunakan sysctl */

2013-03-28T23:05:30Z

Tuning Runtime menggunakan sysctl

Onnowpurbo: /* Tuning Runtime menggunakan sysctl */

2013-03-28T23:00:02Z

Tuning Runtime menggunakan sysctl

@@ Line 441: / Line 441: @@
 * [[Linux]]
 * [[Ubuntu]]
 * [[Linux: CLI untuk Survival]]
-* [[Sistem Operasi]]
+* [[Linux: Skema Partisi di Linux]]
 * [[Linux: Instalasi Sistem Operasi]]
-* [[Linux: Skema Partisi di Linux]]
 * [[Kernel]]
 * [[Compile Kernel]]
 * [[Compile Kernel: Konfigurasi Kernel]]
-* [[Kernel: Anatomi Kernel Source]]
-* [[OS: Linux Kernel]]
+===Remaster Linux===
 * [[OS: Parameter Kernel Default]]
 * [[OS: Kernel Scheduler]]
 * [[OS: Complete Teori Tuning Kernel Scheduler]]
 * [[OS: Complete Teori Tuning I/O Performance]]
 * [[OS: Tuning Manajemen Memory]]
-* [[OS: Tuning Kernel Scheduler]]
-* [[OS: Tuning Completely Fair Queueing CFQ I/O scheduler]]
+===Android===
-* [[OS: Tuning Completely Fair scheduler CFS]]
 * [[OS: Build in Monitoring Tool]]
 * [[Linux Benchmarking]]
 * [[OS: Benchmarking menggunakan UnixBench]]
 * [[OS: Benchmarking menggunakan LLCBench]]

@@ Line 441: / Line 441: @@
 * [[Linux]]
 * [[Ubuntu]]
 * [[Sistem Operasi]]
 * [[Kernel]]
 * [[Compile Kernel]]
 * [[OS: Linux Kernel]]
 * [[OS: Parameter Kernel Default]]
@@ Line 449: / Line 454: @@
 * [[OS: Complete Teori Tuning Kernel Scheduler]]
 * [[OS: Complete Teori Tuning I/O Performance]]
 * [[OS: Tuning Kernel Scheduler]]
 * [[OS: Tuning Completely Fair Queueing CFQ I/O scheduler]]
@@ Line 456: / Line 462: @@
 * [[OS: Benchmarking menggunakan UnixBench]]
 * [[OS: Benchmarking menggunakan LLCBench]]
 * [[OS: Membuat Kernel Modul]]

@@ Line 434: / Line 434: @@
 * http://www.inf.fu-berlin.de/lehre/SS01/OS/Lectures/Lecture08.pdf
 * http://doc.opensuse.org/documentation/html/openSUSE/opensuse-tuning/book.tuning.html
 * http://doc.opensuse.org/documentation/html/openSUSE/opensuse-tuning/cha.tuning.taskscheduler.html

@@ Line 398: / Line 398: @@
   less /proc/PID/sched
-    Displays scheduling information on the process with id PID.
+Akan keluar seperti
-    saturn.example.com:~ # cat /proc/`pidof nautilus`/sched
+ rsync (10485, #threads: 1)
-     nautilus (4009, #threads: 1)
+ ---------------------------------------------------------
-     ---------------------------------------------------------
+ se.exec_start                      :     176371611.333607
-     se.exec_start                      :    2419575150.560531
+ se.vruntime                        :       2627563.353534
-     se.vruntime                        :      54549795.870151
+ se.sum_exec_runtime                :        141516.884079
-     se.sum_exec_runtime                :       4867855.829415
+ se.statistics.wait_start           :             0.000000
-     se.avg_overlap                     :             0.401317
+ se.statistics.sleep_start          :             0.000000
-     se.avg_wakeup                      :             3.247651
+ se.statistics.block_start          :     176371611.333607
-     se.avg_running                     :             0.323432
+ se.statistics.sleep_max            :         60060.014521
-     se.wait_start                      :             0.000000
+ .....
-     se.sleep_start                     :    2419575150.560531
+ se.statistics.nr_wakeups_affine_attempts:              1910947
-      [...]
+ se.statistics.nr_wakeups_passive   :                    0
-     nr_voluntary_switches              :               938552
+ se.statistics.nr_wakeups_idle      :                    0
-     nr_involuntary_switches            :                71872
+ avg_atom                           :             0.038524
-     se.load.weight                     :                 1024
+ avg_per_cpu                        :             0.268959
-     policy                             :                    0
+ nr_switches                        :              3677210
-     prio                               :                  120
+ nr_voluntary_switches              :              3631250
-     clock-delta                        :                  109
+ nr_involuntary_switches            :                45960
-.5. For More Information¶
+==Informasi Lebih Lanjut==
-To get a compact knowledge about Linux kernel task scheduling, you need to explore several information sources. Here are some of them:
+Baca informasi tentang task scheduler di file di bawah /usr/src/linux/Documentation/scheduler
-    For task scheduler System Calls description, see the relevant manual page (for example man 2 sched_setaffinity).
+==Referensi==
-    General information on scheduling is described in Scheduling wiki page.
+* http://www.inf.fu-berlin.de/lehre/SS01/OS/Lectures/Lecture08.pdf
 ==Referensi==

@@ Line 390: / Line 390: @@
 ===/proc/schedstat===
-Displays statistics relevant to the current run queue. Also domain-specific statistics for SMP systems are displayed for all connected processors. Because the output format is not user-friendly, read the contents of /usr/src/linux/Documentation/scheduler/sched-stats.txt for more information.
+Menampilkan statistik relevan terhadap run queue saat itu. Juga domain-spesifik statistik dari sistem SMP untuk semua prosesor yang tersambung. Karena format output tidak terlalu user-friendly, untuk informasi lebih lanjut ada baiknya baca
 /proc/PID/sched
      Displays scheduling information on the process with id PID.

@@ Line 323: / Line 323: @@
 Berisi nilai saat ini dari semua variabel yang dapat di tuning yang akan mempengaruhi perilaku task scheduler, CFS statistik, dan informasi tentang antrian run di semua processor yang tersedia.
-    saturn.example.com:~ # less /proc/sched_debug
+ less /proc/sched_debug
-    cpu#0, 1864.411 MHz
+Hasilnya kira-kira
-    runnable tasks:
+ Sched Debug Version: v0.10, 3.2.0-29-generic #46-Ubuntu
-      task  PID   tree-key    switches prio exec-runtime    sum-exec sum-sleep
+ ktime                                   : 175814713.744739
-    --------------------------------------------------------------------------
+ sched_clk                               : 152754753.141005
-    R cat 16884 54410632.307072    0   120  54410632.307072 13.836804 0.000000
+ cpu_clk                                 : 175814713.743515
 ===/proc/schedstat===

@@ Line 317: / Line 317: @@
 ==Interface Debugging dan Statistik Scheduler==
-CFS comes with a new improved debugging interface, and provides runtime statistics information. Relevant files were added to the /proc file system, which can be examined simply with the cat or less command. A list of the related /proc files follows with their short description:
+CFS digunakan dengan interface debugging yang baru, dan dapat memberikan informasi statiktik runtime. File yang relevan ada di directory /proc, yang dapat di pelajari hanya dengan printah cat atau less. Sedikit tentang file /proc dan penjelasan pendeknya:
 ===/proc/sched_debug===
-Contains the current values of all tunable variables (see Section 14.4.6, “Runtime Tuning with sysctl”) that affect the task scheduler behavior, CFS statistics, and information about the run queue on all available processors.
+Berisi nilai saat ini dari semua variabel yang dapat di tuning yang akan mempengaruhi perilaku task scheduler, CFS statistik, dan informasi tentang antrian run di semua processor yang tersedia.
      saturn.example.com:~ # less /proc/sched_debug

@@ Line 315: / Line 315: @@
 Menaikan nilai ini biasakan akan menaikan performance untuk thread besar pada  SCHED_OTHER dengan mengorbankan latency yang lebih tinggi untuk task yang real-time.
-==Debugging Interface and Scheduler Statistics==
+==Interface Debugging dan Statistik Scheduler==
 CFS comes with a new improved debugging interface, and provides runtime statistics information. Relevant files were added to the /proc file system, which can be examined simply with the cat or less command. A list of the related /proc files follows with their short description:

@@ Line 294: / Line 294: @@
 '''sched_rt_runtime_us'''
-    Quantum allocated to real-time tasks during sched_rt_period_us. Setting to -1 disables RT bandwidth enforcement. By default, RT tasks may consume 95%CPU/sec, thus leaving 5%CPU/sec or 0.05s to be used by SCHED_OTHER tasks.
+Alokasi waktu untuk task real-time (RT) saat sched_rt_period_us. Menset menjadi -1 akan men-disable pelaksanaan waktu real-time (RT).
 '''sched_features'''

@@ Line 281: / Line 281: @@
 Nilai default adalah 5000000 (ns).
-    Settings larger than half of sched_latency_ns will result in zero wake-up preemption and short duty cycle tasks will be unable to compete with CPU hogs effectively.
+'''WARNING:'''
 '''sched_rt_period_us'''
-    Period over which real-time task bandwidth enforcement is measured. The default value is 1000000 (µs).
+Perioda dimana pelaksanaan task real-time di ukur.
 '''sched_rt_runtime_us'''
@@ Line 295: / Line 298: @@
 '''sched_features'''
-    Provides information about specific debugging features.
+Memberikan informasi tentang fitur debugging yang spesifik.
 '''sched_stat_granularity_ns'''
-    Specifies the granularity for collecting task scheduler statistics.
+Menentukan granularitas untuk pengumpulan statistik task scheduler.
 '''sched_nr_migrate'''
-    Controls how many tasks can be moved across processors through migration software interrupts (softirq). If a large number of tasks is created by SCHED_OTHER policy, they will all be run on the same processor. The default value is 32. Increasing this value gives a performance boost to large SCHED_OTHER threads at the expense of increased latencies for real-time tasks.
+Mengatur berapa task yang dapat di pindahkan antar processor melalui interupsi software migrasi (softirq). Jika sebagian besar task dibuat oleh kebijakan SCHED_OTHER, mereka semua biasakan akan jalan di processor yang sama.
 ==Debugging Interface and Scheduler Statistics==