Hi!
I just tested an "apt-get dist-upgrade" on a debian lenny container to upgrade to squeeze and after that the ressource usage is extremely high:
vztest:~# cat /proc/user_beancounters
Version: 2.5
uid resource held maxheld barrier limit failcnt
101: kmemsize 9689201 15546245 50000000 55000000 4
lockedpages 0 109 256 256 30
privvmpages 151894 230561 300000 350000 11
shmpages 151086 208080 256000 256000 6
dummy 0 0 0 0 0
numproc 5 43 240 240 0
physpages 150932 211221 0 9223372036854775807 0
vmguarpages 0 0 33792 9223372036854775807 0
oomguarpages 150932 211221 26112 9223372036854775807 0
numtcpsock 2 12 360 360 0
numflock 0 9 188 206 0
numpty 1 4 16 16 0
numsiginfo 0 9 256 256 0
tcpsndbuf 25432 117120 1720320 2703360 0
tcprcvbuf 32768 342464 1720320 2703360 0
othersockbuf 0 49000 1126080 2097152 0
dgramrcvbuf 0 8456 262144 262144 0
numothersock 0 23 360 360 0
dcachesize 0 0 3409920 3624960 0
numfile 113 801 9312 9312 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
dummy 0 0 0 0 0
numiptent 10 10 128 128 0
As you can see shmpages are at 151086. Before the upgrade shmpages were at 712.
I tried to kill all running processes and to set a lower limit for shared memory (kernel.shmmax and kernel.shmall). It did not change anything.
Here some output that may be useful:
vztest:~# ps aux
USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
root 1 0.0 0.0 8348 772 ? Ss 00:53 0:00 init [2]
root 9562 0.0 0.0 49164 1156 ? Ss 00:53 0:00 /usr/sbin/sshd
root 9705 0.0 0.2 70452 3332 ? Ss 00:56 0:00 sshd: root@pts/0
root 9707 0.0 0.1 17704 1888 pts/0 Ss 00:56 0:00 -bash
root 10163 0.0 0.0 14804 996 pts/0 R+ 01:24 0:00 ps aux
vztest:~# ipcs
------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
------ Semaphore Arrays --------
key semid owner perms nsems
------ Message Queues --------
key msqid owner perms used-bytes messages
Here you can see that shared memory is limited to 40MB:
vztest:~# ipcs -l
------ Shared Memory Limits --------
max number of segments = 4096
max seg size (kbytes) = 10
max total shared memory (kbytes) = 40960
min seg size (bytes) = 1
------ Semaphore Limits --------
max number of arrays = 128
max semaphores per array = 250
max semaphores system wide = 32000
max ops per semop call = 32
semaphore max value = 32767
------ Messages: Limits --------
max queues system wide = 16
max size of message (bytes) = 8192
default max size of queue (bytes) = 16384
Also there is no space used by tmpfs volumes:
vztest:~# df
Filesystem 1K-blocks Used Available Use% Mounted on
/dev/simfs 2560000 1062068 1497932 42% /
tmpfs 600000 0 600000 0% /lib/init/rw
tmpfs 600000 0 600000 0% /dev/shm
To compare the resource usage I created a new container by using the debian-6.0-amd64-minimal template and in this container shmpages are at 16 (even when running more processes).
Do you have any idea what the problem could be? What could be using so much shared memory here?
Greetings,
Michael
Update:
I don't think that it is related to debian or the "apt-get upgrade". In general all resources stay used after stopping processes or even stopping the whole container.
After stopping all containers and waiting for several minutes vzmemcheck and /proc/user_beancounters on the host-system show that the resources are still used.
I'm using the "Openwall GNU/*/Linux (or Owl)" Live-CD with kernel 2.6.18-194.26.1.el5.028stab079.1.owl2.
[Updated on: Tue, 15 February 2011 21:44]
Report message to a moderator