SOLVED: [appended 'lxc.cgroup.cpuset.cpus = 0,1 appended to /etc/pve/lxc/<CTID>.conf to get rid of the error as below in §1, fyi]
Hi,
1§ I am getting the following error inside a ubuntu 16.04 container inside proxmox v4.4:
"ufs: temp directory created at "/tmp/ufs"
nodejs: src/unix/linux-core.c:747: read_times: Assertion `num == numcpus' failed.
Aborted"
2§ This pointed to a limitation as described at https://github.com/Unitech/PM2/issues/1416#issuecomment-153415598:
"the relevant container setting is:
cgroup setting: cpuset.cpus
explained here: https://access.redhat.com/documenta...tml/Resource_Management_Guide/sec-cpuset.html
I hope that helps others as well.. if you are running on virtualized servers, ask the hoster to adjust the values (they probably wont, as that would mean unlimiting your cpu-count in the vm).
P.S. : Similar problems might occur when running in limited docker containers."
3§ A similar issue was discussed in this forum at https://forum.proxmox.com/threads/lxc-cgroup-cpuset-cpus-not-being-respected.25576/ which asks to make custom changes in /etc/pve/lxc/<CTID>.conf
4§ /etc/pve/lxc/<CTID>.conf states:
# cat /etc/pve/lxc/<CTID>.conf
arch: amd64
cores: 1
hostname: hostname.com
memory: 1024
net0: name=eth0,bridge=vmbr0,gw=192.168.200.1,hwaddr=C2:8A:AE:75:17:A2,ip=192.168.200.103/24,type=veth
ostype: ubuntu
rootfs: local-zfs:subvol-103-disk-1,size=100G
swap: 2048
'pct cpusets' command output that specific container is using CPU 1.
5§ The question is how to specify cpuset.cpus cgroup settings in /etc/pve/lxc/>CTID>.conf?
Any input appreciated!
Cheers,
/z
Hi,
1§ I am getting the following error inside a ubuntu 16.04 container inside proxmox v4.4:
"ufs: temp directory created at "/tmp/ufs"
nodejs: src/unix/linux-core.c:747: read_times: Assertion `num == numcpus' failed.
Aborted"
2§ This pointed to a limitation as described at https://github.com/Unitech/PM2/issues/1416#issuecomment-153415598:
"the relevant container setting is:
cgroup setting: cpuset.cpus
explained here: https://access.redhat.com/documenta...tml/Resource_Management_Guide/sec-cpuset.html
I hope that helps others as well.. if you are running on virtualized servers, ask the hoster to adjust the values (they probably wont, as that would mean unlimiting your cpu-count in the vm).
P.S. : Similar problems might occur when running in limited docker containers."
3§ A similar issue was discussed in this forum at https://forum.proxmox.com/threads/lxc-cgroup-cpuset-cpus-not-being-respected.25576/ which asks to make custom changes in /etc/pve/lxc/<CTID>.conf
4§ /etc/pve/lxc/<CTID>.conf states:
# cat /etc/pve/lxc/<CTID>.conf
arch: amd64
cores: 1
hostname: hostname.com
memory: 1024
net0: name=eth0,bridge=vmbr0,gw=192.168.200.1,hwaddr=C2:8A:AE:75:17:A2,ip=192.168.200.103/24,type=veth
ostype: ubuntu
rootfs: local-zfs:subvol-103-disk-1,size=100G
swap: 2048
'pct cpusets' command output that specific container is using CPU 1.
5§ The question is how to specify cpuset.cpus cgroup settings in /etc/pve/lxc/>CTID>.conf?
Any input appreciated!
Cheers,
/z
Last edited:
