Limiting certain processes to CPU % - Linux

Question

I have the following problem: some processes, generated dynamically, have a tendency to eat 100% of CPU. I would like to limit all the process matching some criterion (e.g. process name) to a certain amount of CPU percentage.

The specific problem I'm trying to solve is harnessing folding@home worker processes. The best solution I could think of is a perl script that's executed periodically and uses the cpulimit utility to limit the processes (if you're interested in more details, check this blog post). It works, but it's a hack :/

Any ideas? I would like to leave the handling of processes to the OS :)

---

Thanks again for the suggestions, but we're still missing the point :)

The "slowDown" solution is essentially what the "cpulimit" utility does. I still have to take care about what processes to slow down, kill the "slowDown" process once the worker process is finished and start new ones for new worker processes. It's precisely what I did with the Perl script and a cron job.

The main problem is that I don't know beforehand what processes to limit. They are generated dynamically.

Maybe there's a way to limit all the processes of one user to a certain amount of CPU percentage? I already set up a user for executing the folding@home jobs, hoping that i could limit him with the /etc/security/limits.conf file. But the nearest I could get there is the total CPU time per user...

It would be cool if to have something that enables you to say: "The sum of all CPU % usage of this user's processes cannot exceed 50%". And then let the processes fight for that 50% of CPU regarding to their priorities...

---

Guys, thanks for your suggestions, but it's not about priorities - I want to limit the CPU % even when there's plenty of CPU time available. The processes are already low priority, so they don't cause any performance issues.

I would just like to prevent the CPU from running on 100% for extended periods...

Solution 1

I had a slightly similar issue with gzip.

Assuming we want to decrease the CPU of a gzip process:

    gzip backup.tar & sleep 2 & cpulimit --limit 10 -e gzip -z

Options:

• I found sleep useful as the cpulimit sometimes didn't pick up the new gzip process immediately
• --limit 10 limits gzip CPU usage to 10%
• -z automatically closes cpulimit when gzip process finishes

Another option is to run the cpulimit daemon.

Solution 2

I don't remember and dont think there was something like this in the unix scheduler. You need a little program which controls the other process and does the following:

loop
    wait for some time tR
    send SIGSTOP to the process you want to be scheduled
    wait for some time tP
    send SIGCONT to the process.
loopEnd

the ratio tR/tP controls the cpu load.

---

Here is a little proof of concept. "busy" is the program which uses up your cpu time and which you want to be slowed-down by "slowDown":

> cat > busy.c:
    main() { while (1) {}; }

> cc -o busy busy.c
> busy &
> top

Tasks: 192 total,   3 running, 189 sleeping,   0 stopped,   0 zombie
Cpu(s): 76.9% us,  6.6% sy,  0.0% ni, 11.9% id,  4.5% wa,  0.0% hi,  0.0% si
Mem:   6139696k total,  6114488k used,    25208k free,   115760k buffers
Swap:  9765368k total,  1606096k used,  8159272k free,  2620712k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
26539 cg        25   0  2416  292  220 R 90.0  0.0   3:25.79 busy
...

> cat > slowDown
while true; do
 kill -s SIGSTOP $1
 sleep 0.1
 kill -s SIGCONT $1
 sleep 0.1
done

> chmod +x slowDown
> slowDown 26539 &
> top
Tasks: 200 total,   4 running, 192 sleeping,   4 stopped,   0 zombie
Cpu(s): 48.5% us, 19.4% sy,  0.0% ni, 20.2% id,  9.8% wa,  0.2% hi,  2.0% si
Mem:   6139696k total,  6115376k used,    24320k free,    96676k buffers
Swap:  9765368k total,  1606096k used,  8159272k free,  2639796k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
26539 cg        16   0  2416  292  220 T 49.7  0.0   6:00.98 busy
...

---

ok, that script needs some more work (for example, to care for being INTR-upted and let the controlled process continue in case it was stopped at that moment), but you get the point. I would also write that little script in C or similar and compute the cpu ratio from a comand line argument....

regards

Solution 3

I think in Linux there is no solution to cap the cpu usage, but there is an acceptable way to limit any process to a certain CPU usage: http://ubuntuforums.org/showthread.php?t=992706

In case they remove the info, here is again

---

1. INSTALL PACKAGES

2. Install cpulimit package. Code:

   sudo apt-get install cpulimit

3. Install gawk package. Code:

   sudo apt-get install gawk

4. CREATE CPULIMIT DAEMON FILE

Open text editor with root privileges and save bellow daemon script text to new file /usr/bin/cpulimit_daemon.sh

Code:

#!/bin/bash
# ==============================================================
# CPU limit daemon - set PID's max. percentage CPU consumptions
# ==============================================================

# Variables
CPU_LIMIT=20        # Maximum percentage CPU consumption by each PID
DAEMON_INTERVAL=3   # Daemon check interval in seconds
BLACK_PROCESSES_LIST=   # Limit only processes defined in this variable. If variable is empty (default) all violating processes are limited.
WHITE_PROCESSES_LIST=   # Limit all processes except processes defined in this variable. If variable is empty (default) all violating processes are limited.

# Check if one of the variables BLACK_PROCESSES_LIST or WHITE_PROCESSES_LIST is defined.
if [[ -n "$BLACK_PROCESSES_LIST" &&  -n "$WHITE_PROCESSES_LIST" ]] ; then    # If both variables are defined then error is produced.
   echo "At least one or both of the variables BLACK_PROCESSES_LIST or WHITE_PROCESSES_LIST must be empty."
   exit 1
elif [[ -n "$BLACK_PROCESSES_LIST" ]] ; then                                 # If this variable is non-empty then set NEW_PIDS_COMMAND variable to bellow command
   NEW_PIDS_COMMAND="top -b -n1 -c | grep -E '$BLACK_PROCESSES_LIST' | gawk '\$9>CPU_LIMIT {print \$1}' CPU_LIMIT=$CPU_LIMIT"
elif [[ -n "$WHITE_PROCESSES_LIST" ]] ; then                                 # If this variable is non-empty then set NEW_PIDS_COMMAND variable to bellow command
   NEW_PIDS_COMMAND="top -b -n1 -c | gawk 'NR>6' | grep -E -v '$WHITE_PROCESSES_LIST' | gawk '\$9>CPU_LIMIT {print \$1}' CPU_LIMIT=$CPU_LIMIT"
else
   NEW_PIDS_COMMAND="top -b -n1 -c | gawk 'NR>6 && \$9>CPU_LIMIT {print \$1}' CPU_LIMIT=$CPU_LIMIT"
fi

# Search and limit violating PIDs
while sleep $DAEMON_INTERVAL
do
   NEW_PIDS=$(eval "$NEW_PIDS_COMMAND")                                                                    # Violating PIDs
   LIMITED_PIDS=$(ps -eo args | gawk '$1=="cpulimit" {print $3}')                                          # Already limited PIDs
   QUEUE_PIDS=$(comm -23 <(echo "$NEW_PIDS" | sort -u) <(echo "$LIMITED_PIDS" | sort -u) | grep -v '^$')   # PIDs in queue

   for i in $QUEUE_PIDS
   do
       cpulimit -p $i -l $CPU_LIMIT -z &   # Limit new violating processes
   done
done

3. CHANGE VARIABLES TO YOUR ENVIRONMENT NEEDS

CPU_LIMIT Change this variable in above script if you would like to omit CPU consumption for every process to any other percentage then 20%. Please read "If using SMP computer" chapter bellow if you have SMP computer (more then 1 CPU or CPU with more then 1 core).

DAEMON_INTERVAL Change this variable in above script if you would like to have more/less regular checking. Interval is in seconds and default is set to 3 seconds.

BLACK_PROCESS_LIST and WHITE_PROCESSES_LIST Variable BLACK_PROCESSES_LIST limits only specified processes. If variable is empty (default) all violating processes are limited.

Variable WHITE_PROCESSES_LIST limits all processes except processes defined in this variable. If variable is empty (default) all violating processes are limited.

One or both of the variables BLACK_PROCESSES_LIST and WHITE_PROCESSES_LIST has to be empty - it is not logical that both variables are defined.

You can specify multiple processes in one of this two variables using delimiter characters "|" (without double quotes). Sample: if you would like to cpulimit all processes except mysql, firefox and gedit processes set variable: WHITE_PROCESSES_LIST="mysql|firefox|gedit"

4. PROCEDURE TO AUTOMATICALLY START DAEMON AT BOOT TIME

5. Set file permissions for root user: Code:

   sudo chmod 755 /usr/bin/cpulimit_daemon.sh

6. Open text editor with root privileges and save bellow script to new file /etc/init.d/cpulimit

Code:

#!/bin/sh
#
# Script to start CPU limit daemon
#
set -e

case "$1" in
start)
if [ $(ps -eo pid,args | gawk '$3=="/usr/bin/cpulimit_daemon.sh"  {print $1}' | wc -l) -eq 0 ]; then
    nohup /usr/bin/cpulimit_daemon.sh >/dev/null 2>&1 &
    ps -eo pid,args | gawk '$3=="/usr/bin/cpulimit_daemon.sh"  {print}' | wc -l | gawk '{ if ($1 == 1) print " * cpulimit daemon started successfully"; else print " * cpulimit daemon can not be started" }'
else
    echo " * cpulimit daemon can't be started, because it is already running"
fi
;;
stop)
CPULIMIT_DAEMON=$(ps -eo pid,args | gawk '$3=="/usr/bin/cpulimit_daemon.sh"  {print $1}' | wc -l)
CPULIMIT_INSTANCE=$(ps -eo pid,args | gawk '$2=="cpulimit" {print $1}' | wc -l)
CPULIMIT_ALL=$((CPULIMIT_DAEMON + CPULIMIT_INSTANCE))
if [ $CPULIMIT_ALL -gt 0 ]; then
    if [ $CPULIMIT_DAEMON -gt 0 ]; then
        ps -eo pid,args | gawk '$3=="/usr/bin/cpulimit_daemon.sh"  {print $1}' | xargs kill -9   # kill cpulimit daemon
    fi

    if [ $CPULIMIT_INSTANCE -gt 0 ]; then
        ps -eo pid,args | gawk '$2=="cpulimit" {print $1}' | xargs kill -9                    # release cpulimited process to normal priority
    fi
    ps -eo pid,args | gawk '$3=="/usr/bin/cpulimit_daemon.sh"  {print}' | wc -l | gawk '{ if ($1 == 1) print " * cpulimit daemon can not be stopped"; else print " * cpulimit daemon stopped successfully" }'
else
    echo " * cpulimit daemon can't be stopped, because it is not running"
fi
;;
restart)
$0 stop
sleep 3
$0 start
;;
status)
ps -eo pid,args | gawk '$3=="/usr/bin/cpulimit_daemon.sh"  {print}' | wc -l | gawk '{ if ($1 == 1) print " * cpulimit daemon is running"; else print " * cpulimit daemon is not running" }'
;;
esac
exit 0

3. Change file's owner to root:

Code:

sudo chown root:root /etc/init.d/cpulimit

4. Change permissions:

Code:

sudo chmod 755 /etc/init.d/cpulimit

5. Add script to boot-up procedure directories: Code:

   sudo update-rc.d cpulimit defaults

6. Reboot to check if script starts cpulimit daemon at boot time: Code:

   sudo reboot

7. MANUALLY CHECK, STOP, START AND RESTART DAEMON

Note: Daemon and service in this tutorial has equal meaning.

Note: For users using prior to Ubuntu 8.10 (like Ubuntu 8.04 LTS) instead of service command use "sudo /etc/init.d/cpulimit status/start/stop/restart" syntax or install sysvconfig package using command: sudo apt-get install sysvconfig

Check if cpulimit service is running Check command returns: "cpulimit daemon is running" if service is running, or "cpulimit daemon is not running" if service is not running. Code:

sudo service cpulimit status

Start cpulimit service You can manually start cpulimit daemon which will start to omit CPU consumption. Code:

sudo service cpulimit start

Stop cpulimit service Stop command stops cpulimit daemon (so no new process will be limited) and also sets to all existing limited processes to have full access to CPU, just like it was before cpulimit was not running. Code:

sudo service cpulimit stop

Restart cpulimit service If you change some variables settings in /usr/bin/cpulimit_daemon.sh like CPU_LIMIT, DAEMON_INTERVAL, BLACK_PROCESSES_LIST or WHITE_PROCESSES_LIST, then after changing settings you must restart service. Code:

sudo service cpulimit restart

6. CHECK CPU CONSUMPTION WITH OR WITHOUT CPULIMIT DAEMON

Without daemon 1. stop cpulimit daemon (sudo service cpulimit stop) 2. execute CPU intensive tasks in background 3. execute command: top and check for %CPU column Result of %CPU is probably more then 20% for each process.

With daemon turned on 1. start cpulimit daemon (sudo service cpulimit start) 2. execute the same CPU intensive tasks in background 3. execute command: top and check for %CPU column Result of %CPU should be maximum 20% for each process. Note: Don't forget at beginning %CPU can be more then 20%, because daemon has to catch violating process in interval of 3 seconds (set in script by default)

7. IF USING SMP COMPUTER

I have tested this code on Intel dual-core CPU computer - that behaves like SMP computer. Don't forget that top command and also cpulimit by default behaves in Irix mode, where 20% means 20% of one CPU. If there are two CPUs (or dual-core) then total %CPU can be 200%. In top command Irix mode can be turned off with command I (pressing +i when top command is running) and Solaris mode is turned on, where total amount of CPU is divided by number of CPUs, so %CPU can be no more then 100% on any number of CPU computer. Please read more info about top command in top man page (search for I command). Please also read more about how cpulimit is operating on SMP computer in cpulimit official page.

But how does cpulimit daemon operates on SMP computer? Always in Irix mode. So if you would like to spend 20% of CPU power on 2-CPU computer then 40% should be used for CPU_LIMIT variable in cpulimit daemon script.

8. UNINSTALL CPULIMIT DAEMON AND CPULIMIT PROGRAM

If you would like to get rid of cpulimit daemon you can clean up your system by removing cpulimit daemon and uninstalling cpulimit program.

1. Stop cpulimit daemon Code:

   sudo service cpulimit stop # Stop cpulimit daemon and all cpulimited processes

2. Remove daemon from boot-up procedure Code:

   sudo update-rc.d -f cpulimit remove # Remove symbolic links

3. Delete boot-up procedure Code:

   sudo rm /etc/init.d/cpulimit # Delete cpulimit boot-up script

4. Delete cpulimit daemon Code:

   sudo rm /usr/bin/cpulimit_daemon.sh # Delete cpulimit daemon script

5. Uninstall cpulimit program Code:

   sudo apt-get remove cpulimit

6. Uninstall gawk program If you don't need this program for any other script, you can remote it. Code:

   sudo apt-get remove gawk

7. NOTE ABOUT AUTHORS

I have just written daemon for cpulimit (bash scripts above). I am not the author of cpulimit project. If you need more info about cpulimit program, please read official cpulimit web page: http://cpulimit.sourceforge.net/.

Regards, Abcuser

Solution 4

I had a similar problem, and the other solutions presented in the thread don't address it at all. My solution works for me right now, but it is suboptimal, particularly for the cases where the process is owned by root.
My workaround for now is to try very hard to make sure that I don't have any long-running processes owned by root (like have backup be done only as a user)

I just installed the hardware sensors applet for gnome, and set up alarms for high and low temperatures on the CPU, and then set up the following commands for each alarm:

low:

mv /tmp/hogs.txt /tmp/hogs.txt.$$ && cat /tmp/hogs.txt.$$ | xargs -n1 kill -CONT

high:

touch /tmp/hogs.txt && ps -eo pcpu,pid | sort -n -r | head -1 | gawk '{ print $2 }' >> /tmp/hogs.txt && xargs -n1 kill -STOP < /tmp/hogs.txt

The good news is that my computer no longer overheats and crashes. The downside is that terminal processes get disconnected from the terminal when they get stopped, and don't get reconnected when they get the CONT signal. The other thing is that if it was an interactive program that caused the overheating (like a certain web browser plugin!) then it will freeze in the middle of what I'm doing while it waits for the CPU to cool off. It would be nicer to have CPU scaling take care of this at a global level, but the problem is that I only have two selectable settings on my CPU and the slow setting isn't slow enough to prevent overheating.

Just to re-iterate here, this has nothing at all to do with process priority, re-nicing,and obviously nothing to do with stopping jobs that run for a long time. This has to do with preventing CPU utilization from staying at 100% for too long, because the hardware is unable to dissipate the heat quickly enough when running at full capacity (idle CPU generates less heat than a fully loaded CPU).

Some other obvious possibilities that might help are:

Lower the CPU speed overall in the BIOS

Replace the heatsink or re-apply the thermal gel to see if that helps

Clean the heatsink with some compressed air

Replace the CPU fan

[edit] Note: no more overheating at 100% CPU when I disable variable fan speed in the bios (asus p5q pro turbo). With the CPU fully loaded, each core tops out at 49 celcius.

Solution 5

Using cgroups' cpu.shares does nothing that a nice value wouldn't do. It sounds like you want to actually throttle the processes, which can definitely be done.

You will need to use a script or two, and/or edit /etc/cgconfig.conf to define the parameters you want.

Specifically, you want to edit the values cpu.cfs_period_us and cpu.cfs_quota_us. The process will then be allowed to run for cpu.cfs_quota_us microseconds per cpu.cfs_period_us microseconds.

For example:

If cpu.cfs_period_us = 50000 and cpu.cfs_quota_us = 10000 then the process will receive 20% of the CPU time maximum, no matter what else is going on.

In this screenshot I have given the process 2% of CPU time:

As far as the process is concerned it is running at 100%.

Settings cpu.shares on the other hand can and will still use 100% of the idle CPU time.

In this similar example I have given the process cpu.shares = 100 (of 1024):

As you can see the process is still consuming all the idle CPU time.

References:

http://manpages.ubuntu.com/manpages/precise/man5/cgconfig.conf.5.html http://kennystechtalk.blogspot.co.uk/2015/04/throttling-cpu-usage-with-linux-cgroups.html

View more solutions

Comments

[user1568901]

Why is it a bad thing? If they're properly niced, they shouldn't interfere with any other processes. They'll just (in effect) get all the leftover cycles...

[Paul Tomblin]

Reading the blog post, it seems like he wants the CPUs to have some idle time so they don't run so hot. That's why I stopped doing SETI@Home - my CPUs weren't lasting very long because they ran hot all the time.

[user3532201]

Yeah, I've never had a CPU fail due to overheating, despite running SETI@Home. And I never used anything fancier than the retail HSF units that came with my processors.

[Paul]

Perhaps he's in a shared hosting environment and wants to throttle his CPU time so his account doesn't get "locked"

[Nemanja Trifunovic]

I don't know the answer, but maybe you can look at the WINE implementation of SetInformationJobObject Windows API

[Janus Troelsen]

You can use cgroups to achieve this.

[Ken Sharp]

@asparagus Did you try any of the solutions?

[Ciro Santilli OurBigBook.com]

unix.stackexchange.com/questions/151883/…

[dashesy]

Make it a systemd service then systemctl set-property seti.service CPUQuota=50% That will use 50% of total all CPUs.

[user2522201]

ulimit -t is not a good suggestion, this will kill the process when its total execution time has exceeded a certain threshold, definitely not desirable when you just want to throttle a process.

[Ozkan]

RLIMIT_CPU doesn't control the maximum percentage of CPU that can be used; it controls the maximum number of seconds that the process can use. The process will receive SIGXCPU signals as it passes the soft limit and SIGKILL when it crosses the hard limit.

[Janus Troelsen]

That is a C function call. In the terminal you can use ulimit. See bash -c "help ulimit". It uses setrlimit

[Janus Troelsen]

There is a solution to cap the usage, see: bash -c "help ulimit"

[f.ardelian]

Far from robust, but it kind of works. In my case, it would mess up the tty if I was using top, it would say it wasn't running when it was, but otherwise it did the job. Thanks a lot!

[Ross]

ulimit is not a solution to limit %

[Ken Sharp]

I've never seen anything so over-the-top. Use cgroups.

[Horace Heaven]

You can use the process name instead of finding out the PID each time using the -e flag e.g. cpulimit -e [process_name] -l 5

[Jaime]

nice does not control the maximum percentage of CPU that a process can use. The niceness is translated to a process priority and not to % of CPU usage

[Jaime]

The cpulimit command works in this way. It monitors the processes you defined. If a process uses more than the limit, it suspends the process. -- You may check the source code in Github.

[asparagus]

Thanks, I'm glad to still get answers (new & better ones) to a question I posed 11 years ago - feels like a lifetime ago