bash fork: retry: Resource temporarily unavailable

Question

I am trying to run a shell script which will create process using a shell script. I get Resource temporarily unavailable error. how to identify which limit (memory/process/filecount) is creating this problem. Below is my ulimit -a results.

core file size          (blocks, -c) 0
data seg size           (kbytes, -d) unlimited
scheduling priority             (-e) 0
file size               (blocks, -f) unlimited
pending signals                 (-i) 563959
max locked memory       (kbytes, -l) unlimited
max memory size         (kbytes, -m) unlimited
open files                      (-n) 65535
pipe size            (512 bytes, -p) 8
POSIX message queues     (bytes, -q) 819200
real-time priority              (-r) 0
stack size              (kbytes, -s) unlimited
cpu time               (seconds, -t) unlimited
max user processes              (-u) 10000000
virtual memory          (kbytes, -v) unlimited
file locks                      (-x) unlimited

Answer 1

For the case in the comments, where you were not using much memory per thread, you were hitting the cgroup limits. You will find the default to be around 12288, but the value is writable:

$ cat /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.max
12288
$ echo 15000 | sudo tee /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.max
15000
$ cat /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.max
15000

EDIT: For more recent versions of Ubuntu (i.e. 24.04) the location has changed:

$ cat /sys/fs/cgroup/user.slice/user-1000.slice/pids.max
20668 

And if I use my "what is the thread limit" program (found here) to check, before:

$ ./thread-limit
Creating threads ...
100 threads so far ...
200 threads so far ...
...
12100 threads so far ...
12200 threads so far ...
Failed with return code 11 creating thread 12281 (Resource temporarily unavailable).
Malloc worked, hmmm

and after:

$ ./thread-limit
Creating threads ...
100 threads so far ...
200 threads so far ...
300 threads so far ...
...
14700 threads so far ...
14800 threads so far ...
14900 threads so far ...
Failed with return code 11 creating thread 14993 (Resource temporarily unavailable).
Malloc worked, hmmm

Of course, the numbers above are not exact because the "doug" user has a few other threads running, such as my SSH sessions to my sever. Check with:

$ cat /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.current
8

EDIT: For more recent versions of Ubuntu (i.e. 24.04) the location has changed:

$ cat /sys/fs/cgroup/user.slice/user-1000.slice/pids.current
12

Program used:

/* compile with:   gcc -pthread -o thread-limit thread-limit.c */
/* originally from: http://www.volano.com/linuxnotes.html */
#include <stdlib.h>
#include <stdio.h>
#include <unistd.h>
#include <pthread.h>
#include <string.h>
#define MAX_THREADS 100000
#define PTHREAD_STACK_MIN 110241024*1024
int i;
void run(void) {
  sleep(60 * 60);
}
int main(int argc, char *argv[]) {
  int rc = 0;
  pthread_t thread[MAX_THREADS];
  pthread_attr_t thread_attr;
pthread_attr_init(&thread_attr);
  pthread_attr_setstacksize(&thread_attr, PTHREAD_STACK_MIN);
printf("Creating threads ...\n");
  for (i = 0; i < MAX_THREADS && rc == 0; i++) {
    rc = pthread_create(&(thread[i]), &thread_attr, (void *) &run, NULL);
    if (rc == 0) {
      pthread_detach(thread[i]);
      if ((i + 1) % 100 == 0)
    printf("%i threads so far ...\n", i + 1);
    }
    else
    {
      printf("Failed with return code %i creating thread %i (%s).\n",
         rc, i + 1, strerror(rc));
  // can we allocate memory?
  char *block = NULL;
  block = malloc(65545);
  if(block == NULL)
    printf(&quot;Malloc failed too :( \n&quot;);
  else
    printf(&quot;Malloc worked, hmmm\n&quot;);
}

}
sleep(60*60); // ctrl+c to exit; makes it easier to see mem use
  exit(0);
}

See also here

EDIT May, 2020: For newer versions of Ubuntu, the default maximum PID number is now 4194304, and therefore adjusting it is not needed.

Now, if you have enough memory, the next limit will be defined by the default maximum PID number, which is 32768, but is also writable. Obvioulsy in order to have more than 32768 simultaneous processes or tasks or threads their PID will have to be allowed to be higher:

$ cat /proc/sys/kernel/pid_max
32768
$ echo 80000 | sudo tee /proc/sys/kernel/pid_max
80000
$ cat /proc/sys/kernel/pid_max
80000

Note that is quite on purpose that a number bigger than 2**16 was chosen, to see if it was actually allowed. And so now, set the cgroup max to, say 70000:

$ echo 70000 | sudo tee /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.max
70000
$ cat /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.max
70000

And at this point, realize that the above listed program seems to have a limit of about 32768 threads, even if resources are still available, and so use another method. My test server with 16 gigabytes of memory seems to exhaust some other resource at about 62344 tasks, even though there does seem to still be memory available.

$ cat /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.current
62344

top:

top - 13:48:26 up 21:08,  4 users,  load average: 281.52, 134.90, 70.93
Tasks: 62535 total, 201 running, 62334 sleeping,   0 stopped,   0 zombie
%Cpu0  : 96.6 us,  2.4 sy,  0.0 ni,  1.0 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu1  : 95.7 us,  2.4 sy,  0.0 ni,  1.9 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu2  : 95.1 us,  3.1 sy,  0.0 ni,  1.8 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu3  : 93.5 us,  4.6 sy,  0.0 ni,  1.9 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu4  : 94.8 us,  3.4 sy,  0.0 ni,  1.8 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu5  : 95.5 us,  2.6 sy,  0.0 ni,  1.9 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu6  : 94.7 us,  3.5 sy,  0.0 ni,  1.9 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
%Cpu7  : 93.8 us,  4.5 sy,  0.0 ni,  1.7 id,  0.0 wa,  0.0 hi,  0.0 si,  0.0 st
KiB Mem : 15999116 total,   758684 free, 10344908 used,  4895524 buff/cache
KiB Swap: 16472060 total, 16470396 free,     1664 used.  4031160 avail Mem
PID USER      PR  NI    VIRT    RES    SHR S  %CPU %MEM     TIME+ COMMAND
37884 doug      20   0  108052  68920   3104 R   5.7  0.4   1:23.08 top
24075 doug      20   0    4360    652    576 S   0.4  0.0   0:00.31 consume
26006 doug      20   0    4360    796    720 S   0.4  0.0   0:00.09 consume
30062 doug      20   0    4360    732    656 S   0.4  0.0   0:00.17 consume
21009 doug      20   0    4360    748    672 S   0.3  0.0   0:00.26 consume

Seems I finally hit my default ulimit settings for both user processes and number of timers (signals):

$ ulimit -i
62340
doug@s15:~$ ulimit -a
core file size          (blocks, -c) 0
data seg size           (kbytes, -d) unlimited
scheduling priority             (-e) 0
file size               (blocks, -f) unlimited
pending signals                 (-i) 62340
max locked memory       (kbytes, -l) 64
max memory size         (kbytes, -m) unlimited
open files                      (-n) 32768
pipe size            (512 bytes, -p) 8
POSIX message queues     (bytes, -q) 819200
real-time priority              (-r) 0
stack size              (kbytes, -s) 8192
cpu time               (seconds, -t) unlimited
max user processes              (-u) 62340
virtual memory          (kbytes, -v) unlimited
file locks                      (-x) unlimited

If I raise those limits, in my case, I did it via /etc/security/limits.conf:

# /etc/security/limits.conf
#
# S18 specific edits. 2019.12.24
#       also for a ridiculous number of threads test.
#
#Each line describes a limit for a user in the form:
#
#<domain>        <type>  <item>  <value>
#
#Where:
#<domain> can be:
#        - a user name
#        - a group name, with @group syntax
#        - the wildcard *, for default entry
#        - the wildcard %, can be also used with %group syntax,
#                 for maxlogin limit
#        - NOTE: group and wildcard limits are not applied to root.
#          To apply a limit to the root user, <domain> must be
#          the literal username root.
#
#<type> can have the two values:
#        - "soft" for enforcing the soft limits
#        - "hard" for enforcing hard limits
#
#<item> can be one of the following:
#        - core - limits the core file size (KB)
#        - data - max data size (KB)
#        - fsize - maximum filesize (KB)
#        - memlock - max locked-in-memory address space (KB)
#        - nofile - max number of open file descriptors
* - nofile 32768
#        - rss - max resident set size (KB)
#        - stack - max stack size (KB)
#        - cpu - max CPU time (MIN)
#        - nproc - max number of processes
* - nproc 200000
#        - as - address space limit (KB)
#        - maxlogins - max number of logins for this user
#        - maxsyslogins - max number of logins on the system
#        - priority - the priority to run user process with
#        - locks - max number of file locks the user can hold
#        - sigpending - max number of pending signals
* - sigpending 200000
#        - msgqueue - max memory used by POSIX message queues (bytes)
#        - nice - max nice priority allowed to raise to values: [-20, 19]
#        - rtprio - max realtime priority
#        - chroot - change root to directory (Debian-specific)
#
#<domain>      <type>  <item>         <value>
#
#*               soft    core            0
#root            hard    core            100000
#*               hard    rss             10000
#@student        hard    nproc           20
#@faculty        soft    nproc           20
#@faculty        hard    nproc           50
#ftp             hard    nproc           0
#ftp             -       chroot          /ftp
#@student        -       maxlogins       4
End of file

I am able to go to 126020 threads, before the return of the inability to fork. This time the limit was (keep in mind that there are about `150 root owned threads on this server, before the test starts):

cat /proc/sys/kernel/threads-max
126189

O.K. so now adjusting that parameter:

echo 99999999 | sudo tee /proc/sys/kernel/threads-max
99999999

I can get to about 132,000 threads before my 16 gigabyte server starts to swap memory, and trouble errupts.

$ cat /sys/fs/cgroup/pids/user.slice/user-1000.slice/pids.current
132016

Note: running top places a significant additional load on the system under these conditions, so I didn't run it. However memory:

doug@s18:~/config/etc/security$ free -m
              total        used        free      shared  buff/cache   available
Mem:          15859       15509         270           1          79         137
Swap:          2047           4        2043

At some point you will get into trouble, but it is absolutely amazing how gracefully the system bogs down. Once my system starts to swap, it totally boggs down and I had many of these errors:

Feb 17 16:13:02 s15 kernel: [  967.907305] INFO: task waiter:119371 blocked for more than 120 seconds.
Feb 17 16:13:02 s15 kernel: [  967.907335]       Not tainted 4.10.0-rc8-stock #194
Feb 17 16:13:02 s15 kernel: [  967.907357] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.

And my load average ballooned to ~29000. But I just left the computer for an hour and it sorted itself out. I staggered the spin out of the threads by 200 microseconds per spin out, and that also seemed to help.