rootLab safetyRun these on a disposable VM, not on a production server.
1. A filesystem is 95% full and growing
Task: find what is growing, free space safely and prevent a repeat.
df -hT; df -i du -xh --max-depth=1 /var | sort -h | tail lsof +L1 journalctl --vacuum-size=200M
Check deleted-open files, inodes and log rotation; extend the volume if the data is legitimate. Full walkthrough →
2. A service is not responding after a change
Task: find out why it will not start or answer, without guessing.
systemctl status SERVICE -l --no-pager journalctl -xeu SERVICE --no-pager | tail -40 ss -ltnp SERVICE_BINARY -t # config test where supported
Look at the exit status, port conflicts, permissions, SELinux and dependencies. Full walkthrough →
3. Load average is high but the CPU looks idle
Task: decide whether it is CPU, I/O or memory, and find the process.
uptime; nproc vmstat 1 5 iostat -xz 1 5 ps -eo state,pid,wchan:20,cmd | awk '$1=="D"'
High wa/b and D-state processes point at storage or NFS. Full walkthrough →
4. An application suddenly cannot write: “Read-only file system”
Task: confirm what is read-only and decide whether it is safe to remount.
findmnt -no TARGET,OPTIONS /data dmesg -T | grep -i -E 'remount|I/O error|EXT4-fs|XFS' lsblk -o NAME,RO,MOUNTPOINT
Look for storage errors before repairing. Full walkthrough →
5. You cannot SSH to a server (connection refused)
Task: work out which layer is at fault from another host and from the console.
nc -vz HOST 22 # from the client: refused vs timeout systemctl status sshd # on the server (console) ss -ltnp | grep ':22' firewall-cmd --list-all sshd -t
Refused means a host replied; a timeout usually means a firewall or routing drop. More →
6. Extend /var by 10 GB online
Task: add space to a mounted LVM logical volume without downtime.
vgs; lvs; lsblk vgextend vg01 /dev/sdc # only if the VG has no free space lvextend -r -L +10G /dev/vg01/lv_var df -h /var
Check vgs first. Extending LVM →