Over the last 24 hours I’ve encounted a slightly frustrating issue with LXD. Running LXD 4.4 on Ubuntu 18.04.5.
Three times per second I am getting syslog entries like the below:
Aug 11 09:02:42 server lxd.daemon[2288]: proc_fuse.c: 880: proc_stat_read: cpu4 from /lxc.payload.amp has unexpected cpu time: 783460 in /proc/stat, 786713 in cpuacct.usage_all; unable to determine idle time
Aug 11 09:02:42 server lxd.daemon[2288]: proc_fuse.c: 880: proc_stat_read: cpu5 from /lxc.payload.amp has unexpected cpu time: 770421 in /proc/stat, 773987 in cpuacct.usage_all; unable to determine idle time
Aug 11 09:02:42 server lxd.daemon[2288]: proc_fuse.c: 880: proc_stat_read: cpu6 from /lxc.payload.amp has unexpected cpu time: 777356 in /proc/stat, 784219 in cpuacct.usage_all; unable to determine idle time
Aug 11 09:02:42 server lxd.daemon[2288]: proc_fuse.c: 880: proc_stat_read: cpu7 from /lxc.payload.amp has unexpected cpu time: 776271 in /proc/stat, 784849 in cpuacct.usage_all; unable to determine idle time
This is only happening in relation to one container (a container I created a few days ago, running Ubuntu 20.04.1). Another older container (running Ubuntu 18.04.5) doesn’t have the same messages appearing for it.
This may be related to the issues mentioned here, here and here. It seems that the behaviour is likely benign, but can I at least suppress the log output to avoid my syslog being filled with junk? I tried including a config in /etc/rsyslog.d/ to suppress the messages (:msg,contains,"unable to determine idle time" stop) but that doesn’t appear to work reliably.
