fork: Cannot allocate memory after each command

postcd · 07-10-2016, 03:18 AM

Hello, this is an OpenVZ VPS with Debian Linux.

it is 128MB RAM and 256MB with burst. But after each command i see fork error. Can you please help to discover where the issue lies and how i can remove this error without upgrading RAM which seems to be alot of free? Thank You

Quote:

root@uanmoqfh # w
08:11:21 up 11:54, 0 users, load average: 0.01, 0.01, 0.00
USER TTY FROM LOGIN@ IDLE JCPU PCPU WHAT
-bash: fork: Cannot allocate memory
root@uanmoqfh # free -mt
total used free shared buffers cached
Mem: 256 68 187 0 0 34
-/+ buffers/cache: 34 221
Swap: 256 8 247
Total: 512 76 435
-bash: fork: Cannot allocate memory
root@uanmoqfh # df -h
Filesystem Size Used Avail Use% Mounted on
/dev/simfs 4.9G 540M 4.4G 11% /
tmpfs 128M 0 128M 0% /lib/init/rw
tmpfs 128M 0 128M 0% /dev/shm
-bash: fork: Cannot allocate memory

postcd · 07-10-2016, 04:47 AM

I was adviced to check bean counters because im hitting some of the limits there:
cat /proc/user_beancounters
If yes, if there are fail counts (failcnt), then i need to ask my VPS provider to raise that limit.

sundialsvcs · 07-11-2016, 07:49 AM

Yes, and also, there are only a very-few "official" return codes from fork. It is not entirely consistent how they are used, especially when (as in this case) an external throttling mechanism is being used. It could be argued that [EAGAIN] should have been returned, but you have no control over which one did get returned.

You should also keep in mind that it may or may not be bean_counters that is supplying the limit. There are several limiters out there, and this particular one has been "superseded" although this of course does not mean that it is no longer found to be in-service somewhere.

TB0ne · 07-11-2016, 09:38 AM

Quote:

Originally Posted by sundialsvcs

Yes, and also, there are only a very-few "official" return codes from fork. It is not entirely consistent how they are used, especially when (as in this case) an external throttling mechanism is being used. It could be argued that [EAGAIN] should have been returned, but you have no control over which one did get returned.

You should also keep in mind that it may or may not be bean_counters that is supplying the limit. There are several limiters out there, and this particular one has been "superseded" although this of course does not mean that it is no longer found to be in-service somewhere.

And the OP can also go back and check the OTHER three threads (going back to 2014), where "fork" is mentioned, and advice was given?

From May of this year:
http://www.linuxquestions.org/questi...ee-4175581011/

...which was marked "SOLVED", and also contains numerous references to other threads the OP opened about how to monitor memory/resources.

Or the duplicate threads from 2014:
http://www.linuxquestions.org/questi...er-4175494476/
http://www.linuxquestions.org/questi...er-4175494339/

...in which the same error message is given.

jpollard · 07-14-2016, 07:56 PM

Might check the contents of the bashrc file in use... I suspect this is a message being generated during the creation of the input prompt string.

I'm basing that on the sequence operations shown - the first prompt string looks normal (but were there any errors immediately before that?). The command itself worked just fine - which shows that fork/exec sequence is having no problems. THEN the error messages are shown. To me, that would imply that during the formatting of the PS1 string things get a bit bogus.