[SOLVED] Cgroups memory not showing up correctly (netdata)

[aghadjip]

Hey Folks, Trying to understand why netdata is showing differing data than the container summary page. I keep getting alerts on high cgroup mem usage, but every time i look at the summary page i see its very low.



However, netdata says im using closer to 3.5 gigs and triggering alerts.

lxc-info is unhelpful as it doesnt have memory info

root@manifold:~# lxc-info -n 108
Name: 108
State: RUNNING
PID: 5319
IP: 10.0.10.11
Link: veth108i0
TX bytes: 13.42 GiB
RX bytes: 204.82 GiB
Total bytes: 218.24 GiB

I want to rightsize this container, i was hoping the memory on the containers summary page would do a better job showing me actual usage.

Any ideas?

 

[Matthias.]

I don't know what exactly netdata shows, but might just be the cache.
Can you enter the VM and show the output of free -h?

 

[aghadjip]

hmm, it does look that way
~ λ free -h total used free shared buff/cache available Mem: 4.0G 534M 5.6M 95M 3.5G 3.5G Swap: 4.0G 127M 3.9G

Im surprised that netdata reports this incorrectly but reports the proper used/free on the bare metal
~$ free -h total used free shared buff/cache available Mem: 62Gi 10Gi 4.8Gi 163Mi 47Gi 51Gi Swap: 7.2Gi 139Mi 7.1Gi

ill dig a little deeper here, thanks

 

[aghadjip]

You are correct, it was the cache, Cleared it and netdata is reporting the memory correctly. Ran a memory muncher and both the UI and netdata report the same actual memory.
It is weird that netdata doesnt report this usage for the bare metal, ill open a bug with netdata to understand this more