Azure managed disks: types, data disks, snapshots, resize

You add a data disk to web01, put a file on it, snapshot it, “lose” the file and get it back. Start web01 and check the disks you already have:

1Start web01 and list your disks

Run this:

az vm start --resource-group rg-saat-vms --name web01
az vm boot-diagnostics enable --resource-group rg-saat-vms --name web01
az disk list --resource-group rg-saat-vms --query "[].{name:name, sku:sku.name, sizeGb:diskSizeGB, tier:tier, state:diskState, zone:zones[0]}" -o table

You should see something like this:

Name                                             Sku          SizeGb    Tier    State     Zone
-----------------------------------------------  -----------  --------  ------  --------  ------
web01_OsDisk_1_4c9e1d0b7a2f4e6c9b8a1d2e3f4a5b6c  Premium_LRS  64        P6      Attached  1
win01_OsDisk_1_8f2a6c4e1b3d4f7a9c0e2b4d6f8a0c1e  Premium_LRS  64        P6      Reserved

web01’s disk is Attached because it’s running. win01 is deallocated, so its disk shows Reserved. Boot diagnostics is now on, which you’ll need in the Break It lab.

2Create a 4 GiB data disk and attach it

Run this:

az disk create --resource-group rg-saat-vms --name web01-data01 --size-gb 4 --sku StandardSSD_LRS --zone 1 --tags course=saatvedha-azure lesson=19 -o none
az vm disk attach --resource-group rg-saat-vms --vm-name web01 --name web01-data01 --lun 0 --caching ReadOnly
az vm show --resource-group rg-saat-vms --name web01 --query "storageProfile.dataDisks[].{lun:lun, name:name, caching:caching}" -o table

You should see this:

Lun    Name          Caching
-----  ------------  ---------
0      web01-data01  ReadOnly

--zone 1 matters: web01 lives in zone 1, so its disks must too. B2ats_v2 takes up to 4 data disks; bigger sizes take more.

3Partition, format and mount it

Save this script on your laptop. The last line saves it with Linux line endings, because Windows ones break bash:

$s = @'
set -e
D=$(readlink -f /dev/disk/azure/scsi1/lun0)
parted $D --script mklabel gpt mkpart data ext4 0% 100%
partprobe $D; sleep 2
mkfs.ext4 -q -L data ${D}1
mkdir -p /data
U=$(blkid -s UUID -o value ${D}1)
echo "UUID=$U /data ext4 defaults,nofail 0 2" >> /etc/fstab
mount -a
echo "invoice batch 2026-10-20" > /data/hello.txt
lsblk -o NAME,SIZE,MOUNTPOINT $D
df -h /data
'@
[IO.File]::WriteAllText('C:\azure-lab\disk-setup.sh', $s.Replace("`r`n", "`n"))

Run it on web01:

az vm run-command invoke --resource-group rg-saat-vms --name web01 --command-id RunShellScript --scripts '@C:\azure-lab\disk-setup.sh' --query "value[0].message" -o tsv

You should see something like this:

Enable succeeded: 
[stdout]
NAME   SIZE MOUNTPOINT
sdb      4G 
└─sdb1   4G /data
Filesystem      Size  Used Avail Use% Mounted on
/dev/sdb1       3.9G   24K  3.7G   1% /data

[stderr]

The script finds the disk by LUN, not by name. Your disk might be sdb or sdc, and that’s fine. The fstab line uses the UUID and nofail.

4Take an incremental snapshot

Run this:

az snapshot create --resource-group rg-saat-vms --name snap-web01-data01 --source web01-data01 --incremental true --sku Standard_LRS --tags course=saatvedha-azure lesson=19 -o none
az snapshot show --resource-group rg-saat-vms --name snap-web01-data01 --query "{name:name, incremental:incremental, sku:sku.name, sizeGb:diskSizeGB}" -o table

You should see this:

Name               Incremental    Sku           SizeGb
-----------------  -------------  ------------  --------
snap-web01-data01  True           Standard_LRS  4

5Lose the file, then restore it from the snapshot

Delete the file, as a user would by mistake:

az vm run-command invoke --resource-group rg-saat-vms --name web01 --command-id RunShellScript --scripts 'rm /data/hello.txt; ls /data' --query "value[0].message" -o tsv

Make a new disk from the snapshot and attach it as LUN 1:

az disk create --resource-group rg-saat-vms --name web01-data01-restore --source snap-web01-data01 --sku StandardSSD_LRS --zone 1 --tags course=saatvedha-azure lesson=19 -o none
az vm disk attach --resource-group rg-saat-vms --vm-name web01 --name web01-data01-restore --lun 1

Mount it read-only, copy the file back and unmount:

az vm run-command invoke --resource-group rg-saat-vms --name web01 --command-id RunShellScript --scripts 'mkdir -p /restore; mount -o ro /dev/disk/azure/scsi1/lun1-part1 /restore; cp /restore/hello.txt /data/; umount /restore; cat /data/hello.txt' --query "value[0].message" -o tsv

You should see this:

Enable succeeded: 
[stdout]
invoice batch 2026-10-20

[stderr]

You copied one file back without touching the rest of the live disk. Now detach the restore disk, so two disks with the same UUID aren’t left on one VM:

az vm disk detach --resource-group rg-saat-vms --vm-name web01 --name web01-data01-restore
az disk delete --resource-group rg-saat-vms --name web01-data01-restore --yes

6Find disks nobody is using

This is the query you’ll run in every cost review. Detached disks keep billing:

az disk list --query "[?diskState=='Unattached'].{name:name, group:resourceGroup, sizeGb:diskSizeGB, sku:sku.name}" -o table

If the restore disk delete went through, this list is empty. In a real subscription, it rarely is.

✅ Checkpoint: /data on web01 is mounted from web01-data01, hello.txt is back, and only the snapshot and web01-data01 are left from this lesson.