[Users] Cannot Allocate Memory after suspend / resume
Robert Heinzmann
Robert.Heinzmann at 1und1.de
Mon Jun 15 09:12:17 EDT 2009
Hello,
I'm trying to use the following mechanism to transparently move machines
from one server to another.
.. machine in /vz/ID/private and /vz/ID/root mounted as LVM LV
1) Suspend the machine (dumpfile /vz/ID/ID.dump)
2) Umount Filesystem (/vz/ID)
3) Create DRBD Device of /vz/ID (external Meta Data)
4) Start and mount the DRBD (/vz/ID)
5) resume the Machine
.. Wait for sync to finish
... Other actions
After resume, the machine has this in the logs:
Jun 15 14:27:01 test4 CRON[2268]: PAM unable to
dlopen(/lib/security/pam_unix.so)
Jun 15 14:27:01 test4 CRON[2268]: PAM [dlerror:
/lib/security/pam_unix.so: failed to map segment from shared object:
Cannot allocate memory]
Jun 15 14:27:01 test4 CRON[2268]: PAM adding faulty module:
/lib/security/pam_unix.so
Jun 15 14:27:01 test4 CRON[2268]: PAM unable to
dlopen(/lib/security/pam_env.so)
Jun 15 14:27:01 test4 CRON[2268]: PAM [dlerror:
/lib/security/pam_env.so: failed to map segment from shared object:
Cannot allocate memory]
Jun 15 14:27:01 test4 CRON[2268]: PAM adding faulty module:
/lib/security/pam_env.so
Jun 15 14:27:01 test4 CRON[2268]: PAM unable to
dlopen(/lib/security/pam_limits.so)
Jun 15 14:27:01 test4 CRON[2268]: PAM [dlerror:
/lib/security/pam_limits.so: failed to map segment from shared object:
Cannot allocate memory]
Jun 15 14:27:01 test4 CRON[2268]: PAM adding faulty module:
/lib/security/pam_limits.so
Jun 15 14:27:01 test4 CRON[2270]: PAM unable to
dlopen(/lib/security/pam_unix.so)
But no beancounters are hit:
test4:/# cat /proc/user_beancounters
Version: 2.5
uid resource held maxheld
barrier limit failcnt
36: kmemsize 3243909 5416555
84172801 100000000 0
lockedpages 0 0
8182 8182 0
privvmpages 84088 91852
512000 640000 0
shmpages 13 14
256000 256000 0
dummy 0 0
0 0 0
numproc 35 61
1000 1000 0
physpages 8127 13348
0 9223372036854775807 0
vmguarpages 0 0
512000 9223372036854775807 0
oomguarpages 8127 13348
512000 9223372036854775807 0
numtcpsock 13 15
1200 1200 0
numflock 5 10
1000 1000 0
numpty 2 2
8 8 0
numsiginfo 0 2
1024 1024 0
tcpsndbuf 247992 247992
12000000 24000000 0
tcprcvbuf 212992 0
12000000 24000000 0
othersockbuf 13872 61592
6000000 12000000 0
dgramrcvbuf 0 2576
512000 512000 0
numothersock 21 45
1200 1200 0
dcachesize 142728 253530
12000000 24000000 0
numfile 672 1144
20000 20000 0
dummy 0 0
0 0 0
dummy 0 0
0 0 0
dummy 0 0
0 0 0
numiptent 1245 1245
2000 2000 0
This is a strange behaviour.
Operating System:
Host : Debian Lenny 64 Bit
Guest/ VE : Debian Etch 32 Bit
Host:
HOST:~/deploy# dpkg -l | grep -e openvz -e vzctl
ii drbd8-module-2.6.26-2-openvz-amd64 2:8.3.1-0+2.6.26-15lenny2
RAID 1 over tcp/ip for Linux kernel module
ii libopenvz-bc-perl 0.02-1
Perl access to OpenVZ Beancounter Data
ii linux-headers-2.6-openvz-amd64 2.6.26+17+lenny1
Header files for Linux 2.6-openvz-amd64
ii linux-headers-2.6.26-2-common-openvz 2.6.26-15lenny3
Common header files for Linux 2.6.26-2-openv
ii linux-headers-2.6.26-2-openvz-amd64 2.6.26-15lenny3
Header files for Linux 2.6.26-2-openvz-amd64
ii linux-image-2.6-openvz-amd64 2.6.26+17+lenny1
Linux 2.6 image on AMD64, OpenVZ support
ii linux-image-2.6.26-2-openvz-amd64 2.6.26-15lenny3
Linux 2.6.26 image on AMD64, OpenVZ support
...
ii vzctl 3.0.22-14
server virtualization solution - control too
ii vzquota 3.0.11-1
server virtualization solution - quota tools
How can I debug ?
Does the suspend also contain file handle information ?
Is it possible to move a machine in suspended state to a different
location ?
Any hints appreshiated ....
Regards,
Robert
---
Robert Heinzmann
More information about the Users
mailing list