Killall script should not unmount /var/lib/rancher/k3s/(server|agent)

[brandond]

The best practices docs at https://docs.k3s.io/reference/resource-profiling#preventing-agents-and-workloads-from-interfering-with-the-cluster-datastore recommend putting the server and agent dirs on different filesystems. However, the killall script unmounts everything under /var/lib/rancher/k3s

We should figure out what specifically under /var/lib/rancher/k3s needs to be unmounted, and do that, instead of unmounting everything.

Internal Ref: SURE-7006

[brandond]

@caroline-suse-rancher I believe this was an internal ask, not sure if there is a Jira for it or not.

[endawkins]

Validated on branch master with commit 98262b5 / version 1.30

Environment Details

Infrastructure

• Cloud
• Hosted

Node(s) CPU architecture, OS, and Version:

Linux ip-172-31-3-48 5.14.21-150500.55.44-default #1 SMP PREEMPT_DYNAMIC Mon Jan 15 10:03:40 UTC 2024 (cc7d8b6) x86_64 x86_64 x86_64 GNU/Linux
NAME="SLES"
VERSION="15-SP5"
VERSION_ID="15.5"
PRETTY_NAME="SUSE Linux Enterprise Server 15 SP5"
ID="sles"
ID_LIKE="suse"
ANSI_COLOR="0;32"
CPE_NAME="cpe:/o:suse:sles:15:sp5"
DOCUMENTATION_URL="https://documentation.suse.com/"

Cluster Configuration:

1 server

Config.yaml:

N/A

Additional files

N/A

Testing Steps

1. Copy config.yaml

$ sudo mkdir -p /etc/rancher/k3s && sudo cp config.yaml /etc/rancher/k3s

2. Install k3s
3. Mount server and agent

sudo mount --bind /var/lib/rancher/k3s/server /var/lib/rancher/k3s/server; sudo mount --bind /var/lib/rancher/k3s/agent /var/lib/rancher/k3s/agent

4. Run killall.sh
5. $ cat /proc/mounts /var/lib/rancher/k3s/server; cat /proc/mounts /var/lib/rancher/k3s/agent
6. Validate that server and agent are present in the mounted filesystems

Replication Results:

• k3s version used for replication:

k3s -v
k3s version v1.30.3+k3s1 (f6466040)
go version go1.22.5

cat /proc/mounts | grep /var/lib/rancher/k3s/agent
cat /proc/mounts | grep /var/lib/rancher/k3s/server

No results display even after having run the mount command to mount both server and agent using the command in the steps.

Validation Results:

• k3s version used for validation:

k3s -v
k3s version v1.30.4-rc1+k3s1 (98262b5d)
go version go1.22.5

ec2-user@ip-[IP_ADDRESS]:~> cat /proc/mounts | grep /var/lib/rancher/k3s/server
/dev/nvme0n1p3 /var/lib/rancher/k3s/server xfs rw,relatime,attr2,inode64,logbufs=8,logbsize=32k,noquota 0 0
ec2-user@ip-[IP_ADDRESS]:~> cat /proc/mounts | grep /var/lib/rancher/k3s/agent
/dev/nvme0n1p3 /var/lib/rancher/k3s/agent xfs rw,relatime,attr2,inode64,logbufs=8,logbsize=32k,noquota 0 0

Additional context / logs:

• Scenario 2: Custom data dir /opt/customdir/agent, /opt/customdir/server

Observations

ubuntu@ip-[IP_ADDRESS]:~$ cat /proc/mounts | grep /opt/customdir/server
/dev/root /opt/customdir/server ext4 rw,relatime,discard,errors=remount-ro,commit=30 0 0
ubuntu@ip-1[IP_ADDRESS]:~$ cat /proc/mounts | grep /opt/customdir/agent
/dev/root /opt/customdir/agent ext4 rw,relatime,discard,errors=remount-ro,commit=30 0 0