LPIC-2 command cheat sheet

The commands 201-450 and 202-450 ask for, grouped by LPI topic number. Storage and kernel work is captured on a real spare disk and real modules; the DNS rows are captured from a client that runs no name server, which is the only way to know a service actually answers.

Debian familyUbuntu 26.04 LTS - kernel 7.0.0-30-generic, systemd 259
RPM familyAlmaLinux 10.2 - kernel 6.12.0-211.7.3.el10_2, systemd 257
Shellzsh 5.9 login shell on both - bash 5.3.9 (Ubuntu) / 5.2.26 (AlmaLinux)
Block devicesone 50 GB disk - RAID and LVM work runs on loop devices
Commands24

Search commands

Topic 201 - Linux kernel

lsmod

Loaded modules. The third column is a reference count - non-zero means it cannot be unloaded until its users go first.

Full guide

bash Example session

$lsmod | head -6; echo "---"; lsmod | wc -l; echo "modules loaded (plus one header line)"Module                  Size  Used byqrtr                   53248  2nfnetlink              20480  1cfg80211             1536000  0binfmt_misc            24576  1intel_rapl_msr         20480  0---49modules loaded (plus one header line)

modinfo <mod> · modinfo -p <mod>

Reads the module file on disk without loading it. `-p` lists the parameters it accepts, which is the fastest answer to "what can I tune".

Full guide

bash Example session

$modinfo dummy | grep -E '^(filename|description|license|depends|vermagic)'filename:       /lib/modules/7.0.0-30-generic/kernel/drivers/net/dummy.ko.zstdescription:    Dummy netdevice driver which discards all packets sent to itlicense:        GPLdepends:vermagic:       7.0.0-30-generic SMP preempt mod_unload modversions

modprobe <mod> <param>=<value>

Loads the module AND its dependencies. `insmod` takes a path and resolves nothing, which is why it fails on anything with a dependency.

Full guide
bash Example session
```
$sudo modprobe dummy numdummies=2 && lsmod | grep '^dummy'; echo "--- loaded"dummy                  16384  0--- loaded
```
modinfo -F depends <mod> · depmod -a

Dependencies come from `modules.dep`, which `depmod` writes. Add a module by hand and modprobe will not find it until you rerun depmod.

Full guide
bash Example session
```
$modinfo -F depends dummy; echo "(empty = no dependencies)"; echo "---"; modinfo -F depends bridge (empty = no dependencies)---stp,llc
```
modprobe -r <mod>

Unloads it and anything it pulled in. Interfaces or devices the module created disappear with it.

Full guide
bash Example session
```
$sudo modprobe -r dummy && lsmod | grep -c '^dummy '; echo "0 means it is gone"00 means it is gone
```
sysctl <name> == cat /proc/sys/<name with / for .>

The dots are directory separators. `net.ipv4.ip_forward` IS `/proc/sys/net/ipv4/ip_forward`.

Full guide
bash Example session
```
$sysctl net.ipv4.ip_forward; echo "---"; cat /proc/sys/net/ipv4/ip_forward; echo "--- the same value, two spellings"net.ipv4.ip_forward = 0---0--- the same value, two spellings
```
sysctl -w <name>=<value> (runtime only)

Changes the running kernel and nothing on disk. Lasts until reboot - which is why a setting "keeps resetting itself".

Full guide
bash Example session
```
$sudo sysctl -w net.ipv4.ip_forward=1 | tail -1; cat /proc/sys/net/ipv4/ip_forwardnet.ipv4.ip_forward = 11
```
/etc/sysctl.d/*.conf then sysctl --system

The persistent half. Writing the file alone changes nothing until reboot; `--system` re-reads them all now.

Full guide
bash Example session
```
$ls /etc/sysctl.d/ 2>/dev/null | head -4; echo "--- where a persistent setting belongs"README.sysctl--- where a persistent setting belongs
```

Topics 203/204 - Filesystems and advanced storage

lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINT <disk>

Prove a disk is spare before destroying it: no partitions, no filesystem, no mountpoint - and check root is elsewhere.

Full guide

bash Example session

$lsblk -o NAME,SIZE,TYPE,FSTYPE,MOUNTPOINT /dev/sdb; echo "--- partitions: $(lsblk -no NAME /dev/sdb | tail -n +2 | wc -l), filesystem: [$(lsblk -dno FSTYPE /dev/sdb)], mounted: [$(lsblk -no MOUNTPOINT /dev/sdb | tr -d ' \n')]"NAME SIZE TYPE FSTYPE MOUNTPOINTsdb    5G disk--- partitions: 0, filesystem: [], mounted: []

pvcreate <disk> · pvs · pvdisplay

A bare PV reports `Allocatable NO` and `Total PE 0` - the extent size is decided by the volume group, so that is not an error.

Full guide
bash Example session
```
$sudo pvcreate -f /dev/sdb && sudo pvs /dev/sdb  Physical volume "/dev/sdb" successfully created.  PV         VG Fmt  Attr PSize PFree  /dev/sdb      lvm2 ---  5.00g 5.00g
```
vgcreate <vg> <pv> · vgdisplay <vg>

The VG sets the extent size - 4 MiB by default. Every LV is a whole number of extents, which is why sizes round.

Full guide
bash Example session
```
$sudo vgcreate lpic2vg /dev/sdb && sudo vgs lpic2vg  Volume group "lpic2vg" successfully created  VG      #PV #LV #SN Attr   VSize  VFree  lpic2vg   1   0   0 wz--n- <5.00g <5.00g
```

lvcreate -n <lv> -L <size> <vg>

Reachable at both `/dev/<vg>/<lv>` and `/dev/mapper/<vg>-<lv>`. `lvs` reports the first, `df` the second.

Full guide

bash Example session

$sudo lvcreate -n data -L 2G lpic2vg && sudo lvs lpic2vg  Logical volume "data" created.  LV   VG      Attr       LSize Pool Origin Data%  Meta%  Move Log Cpy%Sync Convert  data lpic2vg -wi-a----- 2.00g

lvextend -L +<size> <lv> (device only)

Grows the block device. `df` still shows the OLD size - stopping here is the commonest LVM mistake.

Full guide

bash Example session

$sudo lvextend -L +1G /dev/lpic2vg/data 2>&1 | tail -1; df -h /mnt/lpic2 | tail -1; echo "--- the volume grew, the filesystem has not yet"  Logical volume lpic2vg/data successfully resized./dev/mapper/lpic2vg-data  2.0G  536K  1.8G   1% /mnt/lpic2--- the volume grew, the filesystem has not yet

resize2fs <lv> (ext) · xfs_growfs <mountpoint> (XFS)

The second half. `lvextend -r` does both at once. XFS takes the MOUNTPOINT, not the device, and cannot be shrunk at all.

Full guide

bash Example session

$sudo resize2fs /dev/lpic2vg/data 2>&1 | tail -1; df -h /mnt/lpic2 | tail -1; echo "--- now the filesystem has caught up, still mounted throughout" /dev/mapper/lpic2vg-data  2.9G  536K  2.8G   1% /mnt/lpic2--- now the filesystem has caught up, still mounted throughout

mdadm --create /dev/md0 --level=1 --raid-devices=2 --spare-devices=1 <devs>

RAID 1 capacity equals ONE device - the mirrors buy redundancy, not space.

Full guide

bash Example session

$sudo mdadm --create /dev/md0 --level=1 --raid-devices=2 --spare-devices=1 $(losetup -a | grep lpic2 | cut -d: -f1 | tr '\n' ' ') --run 2>&1 | tail -2mdadm: Defaulting to version 1.2 metadatamdadm: array /dev/md0 started.

mdadm --fail <md> <dev> · mdadm --detail <md>

The spare is promoted automatically and the array stays `clean`. Watch a real rebuild in /proc/mdstat.

Full guide

bash Example session

$sudo mdadm --fail /dev/md0 $(losetup -a | grep lpic2 | head -1 | cut -d: -f1) 2>&1 | tail -1; sleep 3; sudo mdadm --detail /dev/md0 | grep -E 'State :|Failed Devices|Spare Devices'             State : clean    Failed Devices : 1     Spare Devices : 0

Topic 207 - Domain name server

named-checkconf

Validate named.conf before restarting anything. Silence means valid - it reports problems and nothing else.

Full guide
bash Example session
```
$sudo named-checkconf && echo "named-checkconf: no output means the syntax is valid, exit $?"named-checkconf: no output means the syntax is valid, exit 0
```
named-checkzone <zone> <file>

Echoes the serial it loaded, so it doubles as a check that you remembered to increment it.

Full guide
bash Example session
```
$sudo named-checkzone lpic2.lab /etc/bind/db.lpic2.labzone lpic2.lab/IN: loaded serial 2026082901OK
```

named-checkzone (a failure)

A bad record names the line and the reason, and the zone is `NOT LOADED`. Restarting with a broken zone keeps serving the OLD data instead.

Full guide

bash Example session

$sudo cp /etc/bind/db.lpic2.lab /tmp/db.good && printf 'bad  IN  A   not-an-address\n' | sudo tee -a /etc/bind/db.lpic2.lab >/dev/null && sudo named-checkzone lpic2.lab /etc/bind/db.lpic2.lab 2>&1 | tail -3; echo "--- exit $?"dns_rdata_fromtext: /etc/bind/db.lpic2.lab:13: near 'not-an-address': bad dotted quadzone lpic2.lab/IN: loading from master file /etc/bind/db.lpic2.lab failed: bad dotted quadzone lpic2.lab/IN: not loaded due to errors.--- exit 0

dig @<server> <name> +short (from a CLIENT)

The only query worth trusting. From the server itself it proves the zone parsed and nothing about reachability.

Full guide
bash Example session
```
$dig @192.168.0.78 ns1.lpic2.lab +short; echo "--- THIS is the test that matters"192.168.0.78--- THIS is the test that matters
```

dig @<server> <cname> +short

A CNAME shows the whole chain - target name first, then its address.

Full guide

bash Example session

$dig @192.168.0.78 www.lpic2.lab +short; echo "--- a CNAME resolves to its target and then to an address"web.lpic2.lab.192.168.0.79--- a CNAME resolves to its target and then to an address

dig <name> → status: NXDOMAIN

The name does not exist, authoritatively (`aa` flag). `+short` shows nothing AND exits 0, which is the trap.

Full guide

bash Example session

$dig @192.168.0.78 nothing.lpic2.lab | grep -E 'status:|AUTHORITY SECTION' -A1 | head -4; echo "--- NXDOMAIN, with the SOA saying how long to cache that fact";; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 29365;; flags: qr aa rd; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 1--;; AUTHORITY SECTION:--- NXDOMAIN, with the SOA saying how long to cache that fact

dig <out-of-zone name> → status: REFUSED

Not a failure - a policy. `recursion no` means this server answers only for its own zones, which is what an authoritative server should do.

Full guide
bash Example session
```
$dig @192.168.0.78 www.google.com | grep -E 'status:' | head -1; echo "--- REFUSED rather than a wrong answer";; ->>HEADER<<- opcode: QUERY, status: REFUSED, id: 47785--- REFUSED rather than a wrong answer
```

communications error (nothing listening)

A transport error, not a DNS status - no header, no flags. Distinguish it from a silent timeout, which usually means a firewall dropped the packet.

Full guide

bash Example session

$dig @192.168.0.78 ns1.lpic2.lab +short 2>&1 | head -2; echo "--- nothing answers on .78 any more (exit $?)";; communications error to 192.168.0.78#53: connection refused;; communications error to 192.168.0.78#53: connection refused--- nothing answers on .78 any more (exit 0)

← Back to the LPIC-2 path