AIX 7.1 → 7.2 Migration

Runbook ID

RB-PATCH-AIX-002

Operation

Version migration · 7.1 → 7.2

Method

nimadm (NIM Alt-Disk Migration)

Source level

7100-05-12-2336

Target level

7200-05-10

NIM master

dr_nim_01

Client LPAR

lpar01

Working VG

nimadmvg · ≥ 2× rootvg

Spare disk

hdisk1 · ≥ rootvg

Window

3 — 4 hr

nimadm runtime

90 — 180 min

Cutover

single reboot

NIM Master

dr_nim_01

≥ 7.2 TL5 — must be at or above target

lpp_source · SPOT · nimadmvg working store

──▶

nimadm -c

Client LPAR

lpar01

7.1 TL5 SP12 → 7.2 TL5 SP10

rootvg cloned to hdisk1 · live system untouched

Worked example: All commands assume lpar01 as the client (already a defined NIM machine object on dr_nim_01), hdisk0 = live 7.1 rootvg, hdisk1 = unassigned spare, lpp_source lpp_aix72tl5sp10, SPOT spot_aix72tl5sp10, working VG nimadmvg, media path /export/nim/media/AIX_7.2_TL5_SP10/. Sub in your own values throughout.

Out of scope: new installations / "preservation" installs · PowerHA / HACMP-clustered nodes (cluster must be quiesced; migrate one node at a time per IBM's PowerHA migration procedure) · LPAR migrations (LPM is unrelated to nimadm). For 7.1 → 7.3 the procedure is identical — use a 7.3 NIM master and target a 7.3 lpp_source.

SP version drift: SP versions evolve. Confirm the latest 7.2 TL5 SP available on IBM Entitled Software Support (ESS) and target that — there is no benefit to migrating to an older SP and then patching forward.

Phase 1 Pre-change planning · T-7 to T-2 days

Change ticket raised — CAB approval logged.
Maintenance window agreed with application owners.
AIX 7.2 entitlement confirmed in ESS.
AIX 7.2 base media downloaded to NIM master.
NIM master at ≥ target client level — oslevel -s on master must be ≥ 7200-05-10.
NIM master has free space for lpp_source (~6 GB) and SPOT.
NIM master has nimadmvg with ≥ 2× client rootvg in free PPs.
Client LPAR has free disk for migration target (≥ rootvg size).
Client reachable from NIM master and resolves both ways.
Backup strategy confirmed — mksysb to NIM, plus TSM file-level coverage.
Third-party software confirmed compatible with AIX 7.2 (see note below).
HMC console access tested for the client LPAR.

Third-party software compatibility matters more here than for an SP. Verify against AIX 7.2 TL5:

Component	What to check
TSM / IBM Spectrum Protect client	Per IBM support matrix for AIX 7.2 TL5 SPx supported client versions. Older clients may need upgrading post-migration.
PowerHA / HACMP	Hard version dependency. Do not migrate clustered nodes ad-hoc.
Oracle, DB2, SAP	Each has its own AIX 7.2 certification matrix.
Kernel extensions	`genkex` shows the loaded list — anything 3rd-party must be re-validated.
Java, OpenSSL, OpenSSH	Ship at newer baselines on 7.2.
Monitoring / AV agents	ITM, Nagios, Tanium, etc. — confirm with vendor.

Phase 2 NIM master preparation · T-7 to T-1 days

Skip-ahead: If lpp_source and SPOT already exist for another client at the same target level, jump to step 06.

01 Verify NIM master state and level CRITICAL dr_nim_01 ▶

oslevel -s # Must be >= 7200-05-10 (target client level) or higher. # A 7.1 NIM master CANNOT migrate clients to 7.2. lsnim -l master # Confirm master is initialized. lssrc -ls nimesis # NIM service must be active.

Hard rule: NIM master must be at or above the target client level. A 7.1 master cannot migrate a client to 7.2 — there's no "upgrade" path mid-flight. If the master is below target, upgrade the master first.

02 Stage the AIX 7.2 base media on the master CONFIG dr_nim_01 ▶

The 7.2 install media from ESS comes as one or more ISOs (typically AIX_7200-05-10_DVD_1_of_2.iso and _2_of_2.iso).

mkdir -p /export/nim/media/AIX_7.2_TL5_SP10 cd /export/nim/media/AIX_7.2_TL5_SP10 Transfer ISOs here (scp, sftp, or NFS from your media share) ls -l # Expect: # AIX_7200-05-10_DVD_1_of_2.iso # AIX_7200-05-10_DVD_2_of_2.iso Verify checksums against ESS download page csum -h SHA256 *.iso

✓ Verify: SHA256 from csum matches the value listed on the ESS download page for each ISO. A bad ISO becomes a corrupt lpp_source — catch it at the doorstep.

03 Define the lpp_source from both ISOs CRITICAL dr_nim_01 ▶

The lpp_source is the fileset repository — a directory of .bff files extracted from the install media. ISO 1 builds the initial lpp_source; ISO 2's contents are layered in via bffcreate.

Create the target directory for the lpp_source mkdir -p /export/nim/lpp_source/lpp_aix72tl5sp10 Define the lpp_source from the first ISO nim -o define -t lpp_source \ -a server=master \ -a location=/export/nim/lpp_source/lpp_aix72tl5sp10 \ -a source=/export/nim/media/AIX_7.2_TL5_SP10/AIX_7200-05-10_DVD_1_of_2.iso \ -a packages=all \ lpp_aix72tl5sp10 Add the contents of ISO 2 to the same lpp_source loopmount -i /export/nim/media/AIX_7.2_TL5_SP10/AIX_7200-05-10_DVD_2_of_2.iso \ -o "-V cdrfs -o ro" \ -m /mnt/iso2 bffcreate -d /mnt/iso2/installp/ppc \ -t /export/nim/lpp_source/lpp_aix72tl5sp10/installp/ppc \ all loopumount -i /export/nim/media/AIX_7.2_TL5_SP10/AIX_7200-05-10_DVD_2_of_2.iso \ -m /mnt/iso2 Rebuild the .toc so NIM sees the new filesets cd /export/nim/lpp_source/lpp_aix72tl5sp10/installp/ppc inutoc . Tell NIM to re-check the lpp_source nim -o check lpp_aix72tl5sp10 lsnim -l lpp_aix72tl5sp10 # Look for: simages = yes (means it contains a complete BOS image)

simages = yes is non-negotiable. Without it, nimadm refuses to use this lpp_source for migration. If the attribute is absent, the lpp_source is incomplete — confirm both ISOs were processed and that inutoc ran successfully against the merged tree.

04 Define the SPOT CONFIG dr_nim_01 ▶

The SPOT (Shared Product Object Tree) is a bootable mini-image used by NIM operations. Built from the lpp_source.

nim -o define -t spot \ -a server=master \ -a location=/export/nim/spot \ -a source=lpp_aix72tl5sp10 \ spot_aix72tl5sp10 SPOT creation takes 15 — 30 minutes — it builds a bootable image lsnim -l spot_aix72tl5sp10 nim -o check spot_aix72tl5sp10 # Expect no errors.

Tip: Keep the old 7.1 SPOT around until burn-in is complete — you'll need it for the worst-case bos_inst rollback (see RB.3).

05 Verify or create the working VG (nimadmvg) CONFIG dr_nim_01 ▶

nimadm needs scratch space on the NIM master to NFS-export and manipulate the client's cloned rootvg. Must be a separate VG (not rootvg) with at least 2× the client's rootvg size in free PPs.

lsvg # If 'nimadmvg' does not exist, create it on a dedicated disk: # mkvg -y nimadmvg -s 64 hdiskN lsvg nimadmvg # Note FREE PPs and PP SIZE; multiply for free GB.

Sizing: Free GB = FREE PPs × PP SIZE / 1024. Need ≥ 2× the client rootvg used capacity. Skimping here causes phase 7 failures partway through filest migration — and phase 7 is the longest phase.

06 Verify or define the client NIM machine object CONFIG dr_nim_01 ▶

lsnim -l lpar01 # If "0042-053 lsnim: there is no NIM object named 'lpar01'" # then define it: nim -o define -t standalone \ -a platform=chrp \ -a netboot_kernel=64 \ -a if1="find_net lpar01 0" \ -a cable_type1=N/A \ lpar01 Confirm the client is reachable nim -o lslpp lpar01 | head # If this returns lpp data, client communication is working.

✓ Verify: nim -o lslpp lpar01 returns data from the client. If it errors with 0042-006 or similar, fix bidirectional name resolution and NIM port reachability before continuing.

Phase 3 Pre-flight checks on the client LPAR · T-0

Capture everything: Wrap the whole pre-flight in script /tmp/preflight_$(date +%Y%m%d).log on the client.

07 Capture current OS level and identity BASELINE lpar01 ▶

oslevel -s # expect: 7100-05-12-2336 oslevel -r instfix -i | grep ML uname -a prtconf | head -30

08 Verify fileset health CRITICAL lpar01 ▶

lppchk -v # MUST return clean lppchk -c # checksum verify (optional)

Stop condition: If lppchk -v reports issues, remediate before migration. A migration that inherits a broken 7.1 fileset state will not produce a clean 7.2 — the breakage is carried into the new build, sometimes amplified.

09 Identify obsolete or problem filesets VALIDATE lpar01 ▶

lslpp -L | grep -i obsolete # Any filesets here will be removed by the migration -- # confirm nothing application-critical depends on them. Check for filesets known not to migrate cleanly lslpp -L bos.adt.libm bos.compat.libs perfagent.tools 2>/dev/null

Anything matched in the second command is worth confirming with the application team before migration. perfagent.tools in particular has historically caused post-migration noise.

10 Confirm disk space VALIDATE lpar01 ▶

df -g / /usr /var /opt /tmp /home

Filesystem	Min free	Why
/	100 MB	Bootloader, ODM
/usr	3 GB	7.2 fileset footprint is materially larger than 7.1
/var	1 GB	nimadm and migration logs
/tmp	1 GB	installp scratch
/opt	500 MB	Optional product trees

11 Verify rootvg health and target disk CRITICAL lpar01 ▶

lsvg rootvg lsvg -p rootvg lsvg -l rootvg lspv # Identify the spare disk (e.g. hdisk1) -- must show "None" in # the VG column and be >= the size of hdisk0. bootinfo -s hdisk0 # size in MB bootinfo -s hdisk1 # must be >= hdisk0

✓ Verify: hdisk1 shows None in the VG column · bootinfo -s hdisk1 ≥ bootinfo -s hdisk0 · all LVs in lsvg -l rootvg are open/syncd.

12 Snapshot bootlist and error log BASELINE lpar01 ▶

bootlist -m normal -o > /tmp/bootlist_pre.txt bootlist -m service -o >> /tmp/bootlist_pre.txt errpt | head -30 errpt -a > /tmp/errpt_pre.txt

13 Snapshot running state for post-migration diff BASELINE lpar01 ▶

lssrc -a > /tmp/lssrc_pre.txt netstat -rn > /tmp/routing_pre.txt df -g > /tmp/df_pre.txt no -a > /tmp/no_pre.txt lsattr -El sys0 > /tmp/sys0_pre.txt genkex > /tmp/genkex_pre.txt # loaded kernel extensions

Diff baseline post-migration. Kernel extensions matter especially — anything 3rd-party that doesn't reload on 7.2 is the migration's most common silent-failure mode.

Phase 4 Backups

14 Take a NIM-resident mksysb (preferred) CRITICAL dr_nim_01 ▶

A NIM-resident mksysb gives you the option to recover the LPAR via nim -o bos_inst if both the live rootvg and the migrated clone become unbootable.

nim -o define -t mksysb \ -a server=master \ -a location=/export/nim/mksysb/lpar01_pre72_$(date +%Y%m%d).mksysb \ -a source=lpar01 \ -a mk_image=yes \ lpar01_pre72_$(date +%Y%m%d)

Pull, not push: This pulls a mksysb from the client to the master in one operation. Runtime depends on rootvg size and network speed (allow 30 — 90 min).

15 Take a local mksysb (fallback) FALLBACK lpar01 ▶

If a NIM-resident mksysb is impractical, take a local mksysb to NFS.

df -g /backup mksysb -ipX /backup/$(hostname)_pre72_$(date +%Y%m%d_%H%M).mksysb

16 Verify the spare disk is ready for nimadm VALIDATE lpar01 ▶

lspv | grep hdisk1 # Expect: hdisk1 00fXXXXXXXXXXXXX None

Stale VG metadata: If hdisk1 shows old_rootvg, altinst_rootvg, or another stale VG from a previous operation, clean it first:

Only if necessary — destructive to whatever is on hdisk1 alt_rootvg_op -X old_rootvg # for old_rootvg # OR chpv -C hdisk1 # clears VGDA, USE WITH CARE

Do not pre-clone with alt_disk_copy. nimadm creates its own clone of rootvg on hdisk1 — it is the clone operation. Taking alt_disk_copy on top would just consume another disk for no benefit.

Phase 5 Run the migration

17 Final pre-execution checks on the NIM master VALIDATE dr_nim_01 ▶

Confirm all resources are ready lsnim -l lpp_aix72tl5sp10 | grep simages lsnim -l spot_aix72tl5sp10 lsnim -l lpar01 lsvg nimadmvg | grep "FREE PPs"

✓ Acceptance: lpp_source has simages = yes · SPOT exists with no errors · client object resolves · nimadmvg has ≥ 2× client rootvg in free PPs.

18 Optional: nimadm phase 1 dry run PREVIEW dr_nim_01 ▶

nimadm runs in 12 phases (see step 20). You can preview without committing — phase 1 is the validation phase, it checks all prerequisites without modifying anything on the client. Useful for catching configuration errors before the real run.

nimadm -c lpar01 \ -s spot_aix72tl5sp10 \ -l lpp_aix72tl5sp10 \ -j nimadmvg \ -d hdisk1 \ -P 1 # run only phase 1

-P N: Runs only phase N. Cheap to do, catches misconfigured master/client/lpp_source before you sink hours into a real run.

19 Execute the full migration CRITICAL dr_nim_01 ▶

In a captured session script /var/log/nim/nimadm_lpar01_$(date +%Y%m%d_%H%M).log nimadm -c lpar01 \ -s spot_aix72tl5sp10 \ -l lpp_aix72tl5sp10 \ -j nimadmvg \ -d hdisk1 \ -Y Exit the script session when complete exit

Flag	Purpose
-c	Client (NIM machine object name)
-s	SPOT
-l	lpp_source
-j	VG on NIM master for working storage
-d	Destination disk on the client
-Y	Accept software licenses non-interactively

Total runtime: 90 — 180 min depending on rootvg size, network speed, and NIM master CPU.

20 nimadm phases — what to expect in the log REFERENCE dr_nim_01 ▶

Phase 1

Initialization · prereq check

Phase 2

Client alt_disk_install setup

Phase 3

Clone rootvg hdisk0 → hdisk1 on client

Phase 4

Export client's hdisk1 alt rootvg via NFS to master

Phase 5

Mount alt rootvg on master, configure for migration

Phase 6

Run pre-migration scripts

Phase 7 · longest

Migrate filesets · new BOS installed here

Phase 8

Run post-migration scripts

Phase 9

Bosboot the alt rootvg

Phase 10

Unmount and clean up NFS export

Phase 11

Wake up alt_disk on client · re-import VG metadata

Phase 12

Set client bootlist to alt disk · finalize

Common failure modes: insufficient space (phase 7), broken filesets inherited from the client (phase 1 or 7), NFS export problems (phase 4 — check firewalls and /etc/exports on the master). The error at the top of the failure usually states which phase to restart from with -r or -P N.

21 Monitor progress from the client side MONITOR lpar01 ▶

In a separate session on the client, watch for the alt rootvg appearing and the NFS mount lifecycle.

Watch the alt rootvg appear on hdisk1 while true; do lspv | grep hdisk1; sleep 30; done When phase 4 begins, hdisk1 is NFS-mounted FROM the master df -g | grep nfs When phase 12 completes, bootlist will have been updated bootlist -m normal -o # Expect: hdisk1

22 Confirm completion VALIDATE dr_nim_01 ▶

The nimadm log ends with:

Bootlist is set to boot from disk hdisk1. nimadm: Migration completed successfully.

State at this point: the client is still running 7.1 from hdisk0, but is configured to boot 7.2 from hdisk1 on next reboot. The cutover is the reboot in Phase 6.

Phase 6 Cutover and verification

23 Final pre-reboot checks on the client VALIDATE lpar01 ▶

Confirm bootlist points at the migrated disk bootlist -m normal -o # Expect: hdisk1 Confirm services are still up (we haven't rebooted yet) lssrc -a | grep -i inoperative

24 Application shutdown CAUTION lpar01 ▶

Follow your application shutdown runbook. Database engines, JVMs, and middleware should be stopped cleanly before reboot — orphan transactions on a 7.1 socket that comes up under 7.2 are exactly the kind of subtle bug that takes weeks to chase.

25 Reboot onto the migrated rootvg REBOOT lpar01 ▶

sync; sync; sync shutdown -Fr now

Have an HMC console open and visible during the reboot — first boot off a migrated rootvg is the moment things go sideways and you want to see LED codes immediately, not via the SSH timeout.

26 First boot considerations EXPECTED lpar01 ▶

The first boot off the migrated 7.2 rootvg may:

▸ Take longer than usual while the kernel re-syncs ODM and rebuilds device configuration.
▸ Drop to a console TERM prompt before login — enter vt100 or xterm and continue. Subsequent boots will not prompt.
▸ Generate informational entries in errpt during initial config — review but expect some noise.

Don't panic: Any of the above on the first boot is expected. Repeated occurrences on the second boot are not.

27 Post-reboot verification VALIDATE lpar01 ▶

oslevel -s # Expect: 7200-05-10-XXXX (your chosen target SP) oslevel -r # Expect: 7200-05 instfix -i | grep ML # Expect: All filesets for 7200-05_AIX_ML were found. lppchk -v # MUST return clean errpt | head -30 # Compare against /tmp/errpt_pre.txt df -g # /usr will have grown notably diff /tmp/df_pre.txt <(df -g) | head lssrc -a | grep -i inoperative # Anything that should be active but isn't? bootlist -m normal -o # Expect: hdisk1 (the migrated disk is now your live rootvg) lspv # hdisk0 will show as 'old_rootvg' -- this is your rollback Check kernel extensions reloaded diff /tmp/genkex_pre.txt <(genkex) | head

✓ Acceptance: oslevel at target · lppchk clean · ML found · no inoperative subsystems · bootlist on hdisk1 · hdisk0 shows as old_rootvg · kernel extensions reloaded with no surprising drops.

28 Application validation handoff HANDOFF lpar01 ▶

Hand off to application owners for service validation. Treat the change as in-progress until they sign off. Keep the rollback warm — don't run any of the cleanup steps in Phase 8 yet.

Phase 7 Rollback procedure

Trigger conditions: client fails to boot off the migrated rootvg · lppchk -v reports broken filesets post-migration · new error classes in errpt linked to the migration · application owners report a regression that cannot be remediated in the maintenance window.

RB.1 Rollback while booted on the migrated 7.2 rootvg ROLLBACK lpar01 ▶

Repoint bootlist to the original 7.1 disk (now 'old_rootvg') bootlist -m normal hdisk0 bootlist -m normal -o # verify shutdown -Fr now After reboot oslevel -s # expect: 7100-05-12-2336 lspv # rootvg is back on hdisk0; the migrated 7.2 will show as # altinst_rootvg or similar on hdisk1

RB.2 Rollback if the migrated system will not boot ROLLBACK HMC ▶

Open a vterm to the LPAR.
Power off the LPAR (Operations → Shut Down → Immediate).
Activate the LPAR with profile, in SMS mode.
SMS Main Menu → "5. Select Boot Options" → "1. Select Install/Boot Device" → "5. Hard Drive" → select hdisk0 (the original disk) → "2. Normal Mode Boot".
System boots from hdisk0 (original 7.1 rootvg).

Then follow the RB.1 verification to confirm oslevel -s reports the pre-migration level and the bootlist sticks.

RB.3 Worst case: restore from mksysb via NIM DISASTER dr_nim_01 ▶

If neither disk boots:

nim -o bos_inst \ -a source=mksysb \ -a mksysb=lpar01_pre72_<date> \ -a spot=spot_aix71tl5sp12 \ -a no_client_boot=no \ lpar01

This requires a 7.1 SPOT on the NIM master in addition to the 7.2 one created in step 04 — keep the old 7.1 SPOT around until burn-in is complete.

RB.4 Post-rollback cleanup CLEANUP lpar01 ▶

Once stable on the rolled-back rootvg alt_rootvg_op -X altinst_rootvg # Removes the failed 7.2 alt_rootvg

Preserve for IBM support:

▸ /var/log/nim/nimadm_lpar01_*.log on the master
▸ /var/adm/ras/nim.installp on the client
▸ Any errpt -a output from the failed boot attempts

Phase 8 Cleanup · T+5 to T+14 days, after burn-in

Only after the migrated system has run cleanly and application owners have signed off.

29 Client cleanup — free hdisk0 and archive logs CLEANUP lpar01 ▶

Remove the old 7.1 rootvg clone, freeing hdisk0 alt_rootvg_op -X old_rootvg lspv # hdisk0 should now show "None" in VG column. Optional: extend rootvg onto hdisk0 for mirroring or growth # extendvg rootvg hdisk0 # mirrorvg rootvg hdisk0 # bosboot -ad /dev/hdisk0 # bootlist -m normal hdisk1 hdisk0 Archive change logs for audit tar -cvf /backup/change_$(hostname)_$(date +%Y%m%d).tar \ /tmp/preflight_*.log \ /tmp/errpt_pre.txt /tmp/lssrc_pre.txt /tmp/df_pre.txt \ /tmp/no_pre.txt /tmp/sys0_pre.txt /tmp/genkex_pre.txt

Mirroring is the high-value cleanup option: after burn-in, mirror the new 7.2 rootvg onto hdisk0 so you have redundancy on both disks again. Skip if your storage layer already provides redundancy.

30 NIM master cleanup CLEANUP dr_nim_01 ▶

Remove the pre-migration mksysb (after retention period) nim -o remove lpar01_pre72_<date> rm /export/nim/mksysb/lpar01_pre72_<date>.mksysb Keep lpp_source / SPOT if more clients are pending migration Otherwise: # nim -o remove spot_aix72tl5sp10 # nim -o remove lpp_aix72tl5sp10 # rm -rf /export/nim/lpp_source/lpp_aix72tl5sp10 # rm -rf /export/nim/spot/spot_aix72tl5sp10 Working storage # nimadmvg is cleaned automatically by nimadm on success

Update the change ticket with completion timestamp and close. You earned it.

Appendix A NIM resource quick reference

A.1 NIM commands cheatsheet REFERENCE dr_nim_01 ▶

lsnim # List all NIM objects lsnim -l <object> # Show all attributes of an object lsnim -t lpp_source # List objects of type lpp_source lsnim -t spot # List SPOTs lsnim -t standalone # List client (standalone) machines nim -o check <resource> # Re-validate a resource nim -o remove <object> # Delete a NIM object nim -o reset <client> # Clear stuck NIM state on a client nim -o deallocate -a subclass=all <client> # Free all resources allocated to client nimadm -c <client> -P 1 ... # Run only phase 1 (validation) nimadm -c <client> ... # Full migration nimadm -B -c <client> # "BOS only" mode -- skip non-BOS filesets

A.2 nimadm phase failure quick reference REFERENCE dr_nim_01 ▶

Phase	Common failure	Remediation
1	Client unreachable	Check network, ping, rpcinfo
1	lpp_source missing simages=yes	Re-add ISO 2 contents, `inutoc`, `nim -o check`
1	Client lppchk -v not clean	Fix broken filesets on 7.1 first
3	Spare disk not free / wrong size	Confirm hdisk1 has VG=None and size ≥ hdisk0
4	NFS mount fails	Firewall between client and master, `/etc/exports`
7	"Not enough space in /usr" (alt rootvg)	Increase rootvg size or use larger target disk
7	Fileset prereq missing in lpp_source	Add missing fileset to lpp_source, `inutoc`, restart
9	bosboot fails	Usually disk error; check `errpt` on client
11	Cannot wake alt_disk_install	Manual `alt_rootvg_op -W -d hdisk1` on client

Restart options: If nimadm fails mid-run, it can usually be restarted from a specific phase using -r (resume) or -P N (run from phase N). Read the error message at the top of the failure for the suggested restart.

A.3 Worked example — single command sequence REFERENCE dr_nim_01 ▶

Do not follow this blindly — values (level numbers, hostnames, hdisk names, dates) will change. This is a sequence reference, not a copy-paste recipe.

On NIM master · one-time setup mkdir -p /export/nim/media/AIX_7.2_TL5_SP10 # ... transfer ISOs here ... nim -o define -t lpp_source \ -a server=master \ -a location=/export/nim/lpp_source/lpp_aix72tl5sp10 \ -a source=/export/nim/media/AIX_7.2_TL5_SP10/AIX_7200-05-10_DVD_1_of_2.iso \ -a packages=all \ lpp_aix72tl5sp10 # ... add ISO 2 contents via bffcreate, then inutoc and nim -o check ... nim -o define -t spot \ -a server=master \ -a location=/export/nim/spot \ -a source=lpp_aix72tl5sp10 \ spot_aix72tl5sp10 On client · pre-flight oslevel -s lppchk -v lspv bootlist -m normal -o > /tmp/bootlist_pre.txt errpt -a > /tmp/errpt_pre.txt On NIM master · take pre-migration mksysb nim -o define -t mksysb \ -a server=master \ -a location=/export/nim/mksysb/lpar01_pre72_$(date +%Y%m%d).mksysb \ -a source=lpar01 \ -a mk_image=yes \ lpar01_pre72_$(date +%Y%m%d) On NIM master · run migration script /var/log/nim/nimadm_lpar01_$(date +%Y%m%d_%H%M).log nimadm -c lpar01 -s spot_aix72tl5sp10 -l lpp_aix72tl5sp10 \ -j nimadmvg -d hdisk1 -Y exit On client · reboot and verify shutdown -Fr now # ... wait for reboot ... oslevel -s # expect 7200-05-10-XXXX lppchk -v instfix -i | grep ML errpt | head On client · cleanup after burn-in alt_rootvg_op -X old_rootvg

Quick Reference Command Summary

Command	Purpose	Notes
oslevel -s (master)	Confirm NIM master at ≥ target level	Hard gate — 7.1 master can't migrate to 7.2
nim -o define -t lpp_source	Register the AIX 7.2 fileset repository	Phase 2 step 03 — needs simages=yes
bffcreate -d ... -t ...	Layer ISO 2 filesets into the lpp_source	After ISO 1 has been defined
inutoc <dir>	Build .toc for the lpp_source	Re-run after every directory change
nim -o check <resource>	Re-validate an lpp_source or SPOT	Required after inutoc
nim -o define -t spot	Build the bootable SPOT mini-image	Phase 2 step 04 — 15-30 min
nim -o define -t mksysb -a mk_image=yes	Pull mksysb from client to master	Phase 4 step 14 — preferred backup
nimadm -c <client> -P 1 ...	Phase-1-only validation dry run	Cheap insurance before the real run
nimadm -c ... -s ... -l ... -j ... -d ... -Y	Run the full migration	Phase 5 step 19 — 90-180 min
nimadm -B ...	"BOS only" mode — skip non-BOS filesets	Faster but skips third-party filesets
bootlist -m normal -o	Confirm bootlist after migration	Should show hdisk1 post-Phase 12
oslevel -s (client, post-reboot)	Confirm migration succeeded	Expect 7200-05-10-XXXX
lppchk -v	Verify fileset consistency	MUST be clean pre and post
diff /tmp/genkex_pre.txt <(genkex)	Confirm kernel extensions reloaded	Phase 6 step 27 — silent failure mode
bootlist -m normal hdisk0	Roll back to original 7.1 rootvg	RB.1 — old_rootvg becomes live again
nim -o bos_inst -a source=mksysb ...	Worst-case mksysb restore	RB.3 — needs 7.1 SPOT preserved
alt_rootvg_op -X old_rootvg	Free hdisk0 after burn-in	Phase 8 step 29 — only after sign-off
nim -o remove <object>	Delete NIM objects post-cleanup	Phase 8 step 30
nim -o reset <client>	Clear stuck NIM state on a client	Use if a failed nimadm leaves the client wedged

⚠ Key Notes

The NIM master must be at or above the target client level. A 7.1 master cannot migrate clients to 7.2 — there is no in-flight upgrade path. Upgrade the master first.
simages = yes on the lpp_source is non-negotiable. Without it, nimadm refuses to use the lpp_source. If absent, the lpp_source is incomplete — re-add ISO 2's filesets via bffcreate, run inutoc, and nim -o check.
Always run nimadm -P 1 as a dry run before the real migration. Phase 1 is the validation phase — it catches misconfigured master/client/lpp_source without modifying the client.
Do not pre-clone with alt_disk_copy. nimadm is the clone operation; layering alt_disk_copy on top just consumes another disk for no benefit.
nimadmvg must be a separate VG (not rootvg) on the master, sized ≥ 2× the client's rootvg in free PPs. Skimping causes phase 7 failures partway through fileset migration — which is the longest phase to repeat.
Keep the old 7.1 SPOT around until burn-in is complete. RB.3 (worst-case bos_inst restore) needs a 7.1 SPOT in addition to the 7.2 one you built.
Third-party kernel extensions are the most common silent failure mode. Diff genkex pre and post — anything that doesn't reload on 7.2 needs vendor remediation before you sign off.
First boot off the migrated 7.2 rootvg can take longer than usual, drop to a TERM prompt, or generate informational errpt entries. Expected on the first boot, not on the second.
Treat the change as in-progress until application owners sign off. Do not run the cleanup steps in Phase 8 before that — your rollback (old_rootvg on hdisk0) needs to stay warm.
For 7.1 → 7.3 the procedure is identical — use a 7.3 NIM master and target a 7.3 lpp_source. Substitute throughout.