How to stop syslog message in console ?

[Kan]

I regulary have some errors in host console (about 10 a day)

[1679843.646789] [Hardware Error]: CPU:0 (10:8:1) MC4_STATUS[-|CE|MiscV|-|AddrV|CECC]: 0x9c5a400014080813
[1679843.651169] [Hardware Error]: MC4 Error (node 0): DRAM ECC error detected on the NB.

These errors appear in containers consoles too.

How can I make it 'silent' in containers console but keep them in host console (proxmox 4.4) ?

 

[dietmar]

replace your bad/damaged RAM modules?

 

[LnxBil]

Besides the correct suggestion from @dietmar, ..

In general, you have to reconfigure your logging system (e.g. rsyslog) to not include kernel messages. I don't know if this is a emerg target or not and it needs some tinkering, but it should work to disable the kernel messages or better discard the messages in your container logging system.

 

[Kan]

This error is not critical because of ECC. And hard to identify the bad RAM module.

I tried to change kernel messaging using this command, but it returns an error.

[root@vm ~]# sysctl -w kernel.printk="4 4 3 7"
error: "Read-only file system" setting key "kernel.printk"

Any suggestion ?

 

[Kan]

Besides the correct suggestion from @dietmar, ..

In general, you have to reconfigure your logging system (e.g. rsyslog) to not include kernel messages. I don't know if this is a emerg target or not and it needs some tinkering, but it should work to disable the kernel messages or better discard the messages in your container logging system.

How can I disable kernel messages in container ?

 

[LnxBil]

How can I disable kernel messages in container ?

By reconfiguring your logging service. Please refer to the container operating system manual for that. This is not a PVE subject.

 

[Kan]

By reconfiguring your logging service. Please refer to the container operating system manual for that. This is not a PVE subject.

That's what I did (maybe wrong). I use the sysctl command in container but it returns a "read-only FS" error (see my command in blue). That's why I suppose this is a pve subject. The container is working well except that message in console.

 

[LnxBil]

sysctl is not the logging system, but the kernel tweak setting. In LXC, there is only one kernel, the one from the host, so you cannot tweak stuff there. Depending on your guest/container OS, the logging service is e.g. syslog, rsyslog. syslog-ng, $whatever they use.

There is a ongoing discussion about that here: https://github.com/lxc/lxd/issues/1397
They set this:

echo 1 | sudo tee /proc/sys/kernel/dmesg_restrict

 

[Kan]

Thx but I've got the same error with this command : read-only file system

I'll try kernel.dmesg_restrict = 1 in /etc/sysctl.conf now.

 

[LnxBil]

Did you try to set it on the host and not on the guest?

 

[Kan]

Did you try to set it on the host and not on the guest?

That's what I finally did but I don't like this solution.