initctl continuously takes 100% of CPU

[Buttay]

Many programs are fairly slow to start on my computer, despite it being relatively new (core i5). Suspecting a heavy CPU usage, I started gnome-system-monitor: all processes were displayed at 0% CPU, but the overall load was 1.33; 1.33; 1.34.

Using the "top" command, however, revealed the real CPU use, with the "initctl" process taking 100%CPU, even 1 hour after startup.

I upgraded to roaring ringtail before checking this but the symptoms were the same with quantal quetzal, so there is a fair chance the causes were identical.

Regards

ProblemType: Bug
DistroRelease: Ubuntu 13.04
Package: upstart 1.8-0ubuntu1
ProcVersionSignature: Ubuntu 3.8.0-19.29-generic 3.8.8
Uname: Linux 3.8.0-19-generic x86_64
ApportVersion: 2.9.2-0ubuntu8
Architecture: amd64
CheckboxSubmission: 2deefc5fd2f1f0ae2fdd4bd781248a2a
CheckboxSystem: daed2f3d6643b4a84b4520a2427f8c2b
Date: Sun Apr 28 13:04:05 2013
ExecutablePath: /sbin/initctl
InstallationDate: Installed on 2012-12-15 (133 days ago)
InstallationMedia: Ubuntu 12.10 "Quantal Quetzal" - Release amd64 (20121017.5)
MarkForUpload: True
ProcEnviron:

ProcKernelCmdline: BOOT_IMAGE=/boot/vmlinuz-3.8.0-19-generic root=UUID=fa624f8f-d9d0-4b09-af8b-88b106aaaf5b ro quiet splash vt.handoff=7
SourcePackage: upstart
UpgradeStatus: Upgraded to raring on 2013-04-28 (0 days ago)
UpstartBugCategory: System

[Buttay (cyril-buttay)]

After killing (kill -9, as kill-15 didn't have any effect) initctl, and rebooting the computer, initctl no longer shows up in the "top" command. Therefore, I cannot reproduce the issue on the same computer. Sorry for the noise!

The applications still are pretty slow to start: running libreoffice from a terminal results in more than 1 minute startup, with a freeze of my screen in the meantime (successive startup are fairly instantaneous though). Same for other applications: gnome-sudoku takes 5 seconds for the first startup, less than 1 for any following startup). I know this is not the place to report it, but how can I report this issue, as I don't know which package is involved?

Regards,

Cyril BUTTAY

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Steve Langasek (vorlon)]

Closing as invalid since the problem is not reproducible.

> The applications still are pretty slow to start: running libreoffice
> from a terminal results in more than 1 minute startup, with a
> freeze of my screen in the meantime (successive startup are fairly
> instantaneous though). Same for other applications:
> gnome-sudoku takes 5 seconds for the first startup, less than 1
> for any following startup). I know this is not the place to report
> it, but how can I report this issue, as I don't know which package
> is involved?

That doesn't sound like a package issue to me; it sounds like you just have a slow disk which it takes a long time to load data from.

[Ketil (ketil-froyn)]

I just had this on my 1 week old new laptop running Ubuntu 14.04.1. I ran "strace -s 128 -p <PID>" where PID was the pid of initctl, and the output looked like this (small excerpt only included):

poll([{fd=3, events=POLLIN}], 1, 4198964686) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964682) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964680) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964677) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964675) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964673) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964670) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964668) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964665) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964663) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964660) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964657) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964654) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964650) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964647) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964645) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964643) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964640) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964638) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964636) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964633) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964631) = 1 ([{fd=3, revents=POLLIN}])
poll([{fd=3, events=POLLIN}], 1, 4198964628) = 1 ([{fd=3, revents=POLLIN}])

I looked in /proc/PID/fd, and it looked like this:

dr-x------ 2 root root 0 okt. 21 07:50 .
dr-xr-xr-x 9 root root 0 okt. 20 10:58 ..
lrwx------ 1 root root 64 okt. 21 07:50 0 -> /dev/null
lrwx------ 1 root root 64 okt. 21 07:50 1 -> /dev/pts/14
lrwx------ 1 root root 64 okt. 21 07:50 2 -> /dev/pts/14
lrwx------ 1 root root 64 okt. 21 07:50 3 -> socket:[10190]

I also ran "lsof | grep 10190", and the output looked like this:

initctl 1268 root 3u unix 0xffff8800d719d400 0t0 10190 socket

Not sure any of this helps, but that's what it is showing while using 100% cpu. My laptop is a Toshiba Portege Z30 with a dual core i7-4510U CPU. I'm running arch amd64.

[Raphael Raccuia (blindekinder)]

any news? happens now often on Toshiba Tecra A50-A i7 - 14.04 64bits - kernel 3.17.0-rc7

[Colin Slaughter (cslaughter-o)]

I would also like to know if anyone has figured this out. I have a Toshiba Terca z50-A i7vPro 14.04 64Bit. It has this problem ever now and again.

[John-William Trenholm (jtrenholm)]

Any update?

[Sander (sander-ubuntu)]

Same issue here, Ubuntu 14.04 64 bits on a Toshiba Portege Z30-A 18J, custom kernel 3.18.11 64 bits though. I suspect it has something to do with certain processes running on the system, such as perhaps Nautilus, but no hard evidence. The problem only occurs after a long uptime duration.
Strace gives the same result as #5 mentioned.

[VF (fviktor)]

I confirm this issue on Ubuntu 14.04 64 bits after approx. 24 hours runtime.

[Юрий Чудновский (fqc)]

Confirming for Ubuntu 15.10 64bit, HP Probook 4320s, after day-two uptime.

[perja (pjagmalm)]

Can confirm this issue on Ubuntu 14.04 & 15.10 64bit, Toshiba Protege Z30, varying uptimes.

[Felix Moreno (info-justdust)]

It's happening me to me in ubuntu 16.10 after a day of uptime maybe...

[Donjan Rodic (bryonak)]

Happens on Ubuntu 16.10 (fresh install, Compiz fallback session) after a few hours of uptime.
htop reports 100% usage of one CPU for the command

/sbin/initctl emit indicator-services-start

which goes on indefinitely. Tracing (with only this initctl process around) yields similar to #5

$ sudo strace -p `pidof initctl`
poll([{fd=3, events=POLLIN}], 1, -1) = 1 ([{fd=3, revents=POLLIN}])
[endless repeat]

/var/log/upstart is empty and dmesg/kern.log/syslog don't help.

The random battery drain is annoying (especially during presentations), and apt purging upstart depends on a little too much for comfort.

Suggestions?

[Arul (arulselvan)]

Anyone found a fix or workaround for this problem? I have the same w/ Ubuntu 16.10

/sbin/initctl emit indicator-services-start

The initctl spinning out of control. This was NOT the case with the previous version 16.04 and it only started to happen with 16.10.

[john wadsworth (jaydub385)]

bump for #15. I am reporting the same issue on my machine.

[Dattée David (dadat)]

bump
Having the same problem too with /sbin/initctl emit indicator-services-start

[Arul (arulselvan)]

Any update on a fix to this problem?

My work around is to kill the initctl if/when I restart; I haven't noticed any adverse effect on anything I do on a daily basis and it is running more than 2+ weeks with no sign of initctl going crazy. Since I very rarely reboot (only after any updates and such), it is not a big deal for me now but would be nice to get a fix to this problem.

Not sure the following info helps but I am sure it will give some clue. I attached gdb and dumped the stack trace after a fresh reboot (i.e. when initctl is fine) as well as the same when it goes crazy to compare. Here are the two traces.

stack trace normal (siting in poll() normally)
==================
(gdb) where
#0 0x00007f00ff2ba0a0 in __poll_nocancel () at ../sysdeps/unix/syscall-template.S:84
#1 0x00007f00ff5ae78f in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#2 0x00007f00ff5ad3fe in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#3 0x00007f00ff596164 in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#4 0x00007f00ff596c5c in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#5 0x0000559a959932c9 in ?? ()
#6 0x00007f00ff9e9ef0 in ?? () from /lib/x86_64-linux-gnu/libnih.so.1
#7 0x00007f00ff9ea513 in nih_command_parser () from /lib/x86_64-linux-gnu/libnih.so.1
#8 0x0000559a95990699 in ?? ()
#9 0x00007f00ff1de3f1 in __libc_start_main (main=0x559a95990620, argc=3,
    argv=0x7fff0ace82d8, init=<optimized out>, fini=<optimized out>,
    rtld_fini=<optimized out>, stack_end=0x7fff0ace82c8) at ../csu/libc-start.c:291
#10 0x0000559a959906ea in ?? ()

stack trace w/ problem (appear to read something from FD 3 (a unix socket to upstart?)
======================
(gdb) where
#0 0x00007f7750af142d in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#1 0x00007f7750af5f41 in dbus_message_get_reply_serial () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#2 0x00007f7750aeb146 in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#3 0x00007f7750aebc5c in ?? () from /lib/x86_64-linux-gnu/libdbus-1.so.3
#4 0x0000557f7db2d2c9 in ?? ()
#5 0x00007f7750f3eef0 in ?? () from /lib/x86_64-linux-gnu/libnih.so.1
#6 0x00007f7750f3f513 in nih_command_parser () from /lib/x86_64-linux-gnu/libnih.so.1
#7 0x0000557f7db2a699 in ?? ()
#8 0x00007f77507333f1 in __libc_start_main (main=0x557f7db2a620, argc=3, argv=0x7fff51adf6d8, init=<optimized out>, fini=<optimized out>,
    rtld_fini=<optimized out>, stack_end=0x7fff51adf6c8) at ../csu/libc-start.c:291
#9 0x0000557f7db2a6ea in ?? ()

[Ketil (ketil-froyn)]

@Arul, I found something similar in my comment above, at https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/1173915/comments/5.

But it's been almost 4 years since this issue was reported, and over 2 years since I added my comment, and it's clearly not fixed yet, or even prioritised.

In fact, my general experience reporting Ubuntu issues here on launchpad is that this is a black hole, issues are left hanging until they are closed as irrelevant several years later. Critical issues undoubtedly get more attention, but issues such as this are still a problem. Annoying, so I don't really see the point in reporting issues here any more.

[Cecil Carpenter (cscj01)]

I have this same problem after a clean upgrade to 16.04 x64 with Intel i5-3570K on a Gigabyte Z77X-UD5H MB with 32 GB of memory. This never occurred on 14.04 on the same machine.

[dw1 (dw1)]

Same on fresh upgrade to 16.04 on Acer E1-531

[Andrew (coder27)]

This happens daily on my 14.04/64 bit

[Arul (arulselvan)]

Just upgraded to 17.04 and this appears to be fixed i.e. I don't see the stupid 'initctl' running anymore!

[pavel heimlich (tropikhajma)]

I was bitten by this on 16.10

[Donjan Rodic (bryonak)]

Realised that this was happening on a few 16.04 LTS machines I manage. It's one thing for me to kill the daemon periodically myself... and completely another to have annoyed users for months without being handled.

For now we settle with crontab -e
*/5 * * * * killall initctl

Killing initctl every 5 minutes looks really scary and it's hard to believe that this is the currently best solution.

[Peter Connolly (pecer)]

@Arul: I'm still seeing it happen on 16.04. Upgrading to 17 isn't an option for some of us as we rely on third parties that don't support non-LTS releases (e.g., DSE Cassandra)
@Ketil: Couldn't agree more. Lower priority bugs are the red-headed stepchild on launchpad.

[Jack Y. Araz (jackaraz)]

Have the same issue

 1504 root 20 0 103700 87464 1752 R 97.1 0.3 4525:02 initctl

> sudo strace -p `pidof initctl`
$ poll([{fd=3, events=POLLIN}], 1, 3729760780) = 1 ([{fd=3, revents=POLLIN}])

endless repeat...

This report is open since 2013 and I'm reporting in 2019. Is there any solution to this issue??

Thanks

[Peter Connolly (pecer)]

Issue finally went away for me after upgrading to 18.04.3 LTS.