IBM DS8000 DSCLI Commands with Examples

This page collects the DS8000 DSCLI commands storage admins use every week on IBM DS8000 systems, with a plain-English line and a working example for each. The syntax follows IBM’s DS CLI reference for the DS8000 G10 family (DS8A10, DS8A50, DS8A80, Release 10.1). Almost all of it also works on a DS8900F.

Use the badges before you press Enter. Read only commands only look. Changes config commands change the setup but don’t destroy data. Danger commands can delete data or cut a host off its disks, so check twice, and check with your change process first.

Want to practise without a real DS8000? Our IBM DS8000 Implementation course gives you a practice shell that runs most of these commands on your laptop, and the IBM DS8000 Architecture course explains what every object really is.

Try it yourself: the DS8000 tab in the Command Builder fills in these commands with your own pool, volume, host and WWPN values: 35 tasks, from a health check to Metro Mirror.

On this page: DS CLI basics · system and hardware · I/O ports · arrays and ranks · extent pools · FB volumes · hosts and mapping · CKD volumes · FlashCopy · Metro Mirror and Global Mirror · Safeguarded Copy capacity · Easy Tier · users · licences · monitoring · recipes · commands that don’t exist

Before you start: how the DS CLI works

The DS CLI runs on your laptop or a jump server and talks to the DS8000 through its HMC (Hardware Management Console). Every command has the same shape: a name, some flags that start with a hyphen, and usually an object ID at the end, for example lsrank -l R0 R1. The names follow a pattern: ls lists, show gives the detail of one object, mk makes, ch changes, rm removes.

A tip from real teams: put devid, hmc1 and hmc2 in your profile. Then you never send a command to the wrong box by forgetting -dev, and you keep working when one HMC is down.

The order you configure a DS8000 in

Every DS8000 is built bottom up, and each step needs the one before it. Array sites already exist when the system arrives (the system makes them from the installed drives), so your work starts with picking one. For open systems (FB) the LSS comes into being by itself when you create the first volume in it. For IBM Z (CKD) you create the LCU yourself before any volume.

1. Array site: pick a free one and note its DA pair. lsarraysite -l
2. Array: RAID 6 or RAID 10 from that site. mkarray -raidtype 6 -arsite S1
3. Rank: format the array for FB or CKD. mkrank -array A0 -stgtype fb or -stgtype ckd
4. Extent pool: one per rank group (server), then put the rank in it. mkextpool -rankgrp 0 -stgtype fb pool_0 and chrank -extpool P0 R0
5. LSS or LCU: FB volumes create their LSS automatically. CKD needs mklcu -qty 1 -id 40 -ss 4040 first.
6. Volumes: mkfbvol -extpool P0 -cap 100 2000-2003 for FB, or mkckdvol -extpool P2 -cap 10017 4000-4003 for CKD, plus mkaliasvol for PAV aliases.
7. Host access: FB volumes are mapped to hosts (mkhost and chhost -action map on DS8900F and DS8A00, or mkvolgrp and mkhostconnect on older systems). CKD volumes go to the FICON/ESCON ALL volume group by themselves, and the z/OS team brings them online.

The Build in order task in the Command Builder writes this whole sequence for you, for FB or CKD.

System and hardware

Run these first when you log in to a box you don’t know, or when someone says “the storage is down”.

I/O ports

When a host complains about slow or flapping paths, showioport -metrics is the fastest check you have. Growing CRCErr or InvTxWordErr counters usually mean a dirty connector, a bad cable or a failing SFP on the switch side.

Arrays and ranks

Extent pools

Extent pools come in pairs: an even pool on rank group 0 (server0) and an odd pool on rank group 1 (server1). Keep the pair balanced, or one server does most of the work.

FB volumes (open systems LUNs)

Volume IDs are four hex digits, LLVV: LL is the LSS (00–FE), VV is the volume inside it (00–FF). Even LSSs come from even pools, odd LSSs from odd pools. The first digit is the address group, which is either all FB or all CKD.

Hosts and volume mapping

The modern way is host objects: create the host with its WWPNs, then map volumes to the host or to a cluster of hosts. The DS8000 creates the volume group for you.

CKD volumes for IBM Z

For mainframes you create LCUs first, then CKD volumes inside them. There are no host objects: FICON access follows the IODF on the IBM Z side.

FlashCopy

FlashCopy makes a point-in-time copy inside the same DS8000. The pair is written source:target, and the target must be the same size or bigger.

Metro Mirror, Global Copy and Global Mirror

Remote copy needs paths first (per LSS pair), then pairs. Metro Mirror (-type mmir) is synchronous; Global Copy (-type gcp) is asynchronous. Global Mirror adds consistency groups on top of Global Copy with a session.

The usual cause of a suspended mirror is not the mirror at all. It’s the paths: both paths on one fabric, a changed zone, or an ISL that went down. Check lspprcpath -l before anything else.

Safeguarded Copy capacity

Be careful with what the DS CLI can and can’t do here. The DS CLI only sets up the backup capacity for Safeguarded Copy. Creating backups, scheduling them, expiring and recovering them is done in IBM Copy Services Manager (CSM) or GDPS, not in the DS CLI.

Easy Tier

Users and security

Licences

Monitoring and support

Three everyday recipes

1. Give a new Linux server four 100 GiB LUNs

Spread the volumes over both servers: two in even LSS 20 (pool P0) and two in odd LSS 21 (pool P1). Then create the host and map them.

lsextpool
mkfbvol -extpool P0 -cap 100 -sam ese -name dbsrv01_#h 2010-2011
mkfbvol -extpool P1 -cap 100 -sam ese -name dbsrv01_#h 2110-2111
lshostport -unknown
mkhost -type "Linux Server" -hostport 10000090FA13B914,10000090FA13B915 dbsrv01
chhost -action map -volume 2010-2011,2110-2111 dbsrv01
lshost -l dbsrv01

On the server, rescan the HBAs and check that every LUN shows at least two paths in multipath -ll before you hand it over.

2. A two-minute health check

lsserver -l
lsioport
lsextpool
lspprc -l 2000-21FF
lsflash 2000-21FF

Both servers Online, the expected ports with Communication established, no pool near its threshold, mirrors in Full Duplex. If any of these look wrong, start there.

3. Before you delete a volume

showfbvol 2011
lspprc 2011
lsflash 2011
lshost -l
chhost -action unmap -volume 2011 dbsrv01
rmfbvol -safe 2011

Check the name and size, check for mirror and FlashCopy relationships, confirm the host owner has stopped using it, unmap it, wait for your agreed cooling-off period, then delete with -safe. Never jump straight to -force.

Commands people look for that don’t exist

Old notes, forum posts and AI answers often mention DS CLI commands that aren’t in the Release 10.1 reference. Here is what to use instead.

Common questions

Do I need the -dev flag on every command? Only if your profile has no devid. Set it once in the profile and you can leave -dev out, except for commands aimed at the remote box in Copy Services.

Can I run DS CLI commands from the GUI? Yes. The DS GUI has an embedded DS CLI. A few commands, such as offloadfile, applykey and mksyslogserver, don’t run there.

Why does mkfbvol fail in my pool? The usual reasons are an odd LSS in an even pool (or the other way round), an address group already used for CKD, a full pool, or a standard volume in a FlashCore Module pool, which only accepts -sam ese.

Which DS CLI version should I install? One current version usually works with every code level. If you see odd errors, install the version that matches the code bundle shown by ver -l. Code bundles 10.11 and later need Java 21.

Further reading: DS8000 DSCLI commands

IBM’s official guide for this family, including the DS CLI chapter: IBM Storage DS8A00 Architecture and Implementation Guide (IBM Redbooks). To practise every command on this page step by step, take our IBM DS8000 Implementation course.