Monitoring tmpfs

[adamwms (Adam)]

CMK version: 2.1.0p19 CRE

Client OS version: Debian 10.13

I’m trying to monitor /run tmpfs file system.

I’ve tried:

Setup → Discovery rules → Filesystem discovery → Filesystem types to ignore → unticked “tmpfs” → Save → Activate on selected sites

as well as:

Mountpoints to never ignore → Explicit match → “/run” → Save → Activate on selected sites

I’ve tried both options alone and combined together.

In all cases when I go to host in question → Service configuration → Rescan

additional checks for the /run volume don’t appear on the list.

What am I missing?

[aeckstein (Andre Eckstein)]

Hi,

i think that this is explicitly excluded in the linux agent /usr/bin/check_mk_agent itself:

    # The exclusion list is getting a bit of a problem.
    # -l should hide any remote FS but seems to be all but working.
    excludefs="-x smbfs -x cifs -x iso9660 -x udf -x nfsv4 -x nfs -x mvfs -x prl_fs -x squashfs -x devtmpfs -x autofs -x beegfs"
    if [ -z "${IS_LXC_CONTAINER}" ]; then
        excludefs="${excludefs} -x zfs"
    fi

So try to remove the exluded filesystem type there.

[adamwms (Adam)]

I’ve removed “-x devtmpfs” from the list and restarted the daemon “sudo systemctl restart cmk-agent-ctl-daemon.service”.
Unfortunately a service rescan hasn’t returned anything new.

[aeckstein (Andre Eckstein)]

Can you check the if the filesystem is present in the agent output ?
On the monitored linux server:

check_mk_agent | grep -A10 "^<<<df"

[adamwms (Adam)]

It is:

check_mk_agent | grep -A10 "^<<<df"
<<<df>>>
udev                  devtmpfs      475688       0    475688       0% /dev
tmpfs                 tmpfs          98360   16988     81372      18% /run
/dev/mapper/vg00-lv01 ext4        18071748 4843296  12395216      29% /
tmpfs                 tmpfs         491796       4    491792       1% /dev/shm
tmpfs                 tmpfs           5120       0      5120       0% /run/lock
tmpfs                 tmpfs         491796       0    491796       0% /sys/fs/cgroup
/dev/sda1             ext4          472408   83761    359617      19% /boot
<<<df>>>
[df_inodes_start]
udev                  devtmpfs  118922   354  118568    1% /dev
tmpfs                 tmpfs     122949  2075  120874    2% /run
/dev/mapper/vg00-lv01 ext4     1150560 72483 1078077    7% /
tmpfs                 tmpfs     122949     2  122947    1% /dev/shm
tmpfs                 tmpfs     122949     4  122945    1% /run/lock
tmpfs                 tmpfs     122949    17  122932    1% /sys/fs/cgroup
/dev/sda1             ext4      124928   356  124572    1% /boot
[df_inodes_end]
[df_lsblk_start]

[aeckstein (Andre Eckstein)]

I tested that on a current checkmk 2.2 and the /run partition is showing up as a service.
I configured the ruleset like this :

Make sure, that the rule is above the other rules, that might overwrite that parameter, at least in my default installation there is a rule explicitly for checkmk servers.

[adamwms (Adam)]

It’s working after I added the default “cmk/check_mk_server:yes” Filesystem discovery rule label to the client.
No other client has this label explicitly enabled.
How has service discovery worked until this point?

[aeckstein (Andre Eckstein)]

If no rules match, then the defaults take place. (that should explain, why other stuff works )
Are you sure, that the server you are testing with, does not have the cmk/check_mk_server:yes label ?
You can see that during discovery or in the host status page: