I seem to have better memory than I thought. Worked with EMC storage (on AIX) from 2004 to 2014. Some 554 issues were fixed by EMC (because of bugs in new versions of their software), some we were able to create a workaround for (like mksysb not always working because of Powerpath).
What we did is create a script running at boot-time and at shutdown, running basically "pprootdev on" and "pprootdev fix". And we used a modified version of 'pprootdev' that reorders the devices in AIX's CuAt ODM database, so the disk the system was actually booted from appears first - fixing our mksysb problem. I have submitted my version to EMC support, but as far as I know, they haven't changed it during the time I worked with EMC storage.
We found the issue to be intermittent, occurring only occasionally. If fairly persistent, then EMC suggested turning on AIX's low-level debugger and rebooting the system, to catch more diagnostic output so EMC (or IBM) could work with it; see instructions below if you would want to try that.
I have always felt that EMC made a weird decision in combination with AIX. On AIX any multipathing is typically done hierarchically "below" the level of the hdisk (several other items, like pdisks making up an hdisk). EMC chose to put something "on top of" hdisks that it has to manage on its own, having difficulty making AIX fully aware of it.
I hope you find a workable solution soon; my knowledge is probably too old to be useful nowadays.
If you want to invoke the IBM low-level debugger at next boot:
# bosboot -ad /dev/ipldevice -I
When AIX boots, the low-level debugger pauses the system with a message "Welcome to KDB" and a prompt
************* Welcome to KDB *************
Call gimmeabreak...
Static breakpoint:
.gimmeabreak+000000 tweq r8,r8 r8=0
.gimmeabreak+000004 blr <.kdb_init@AF91_62+0001F4> r3=0
KDB(0)> mw enter_dbg
enter_dbg+000000: 00000000 = 42
enter_dbg+000004: 00000000 = .
KDB(0)> g
At the KDB(0) prompt you enter "mw enter_dbg".
At the next prompt you enter the number 42 and press Enter.
At the next prompt you enter a period (.) and press Enter.
At the new KDB(0) prompt you enter "g" (for go) and press Enter.
If the system boots and you want to "undo" the debugging, use bosboot again, without the -I option and reboot the system.
# bosboot -ad /dev/ipldevice
------------------------------
Richard Westerik
Principal specialist
Simac IT NL bv
Ede
+31651575123
------------------------------
Original Message:
Sent: Mon April 20, 2026 03:09 AM
From: Richard Westerik
Subject: Error 554 with DELL (EMC) SAN Storage boot disk.
Hello Luis,
I have been working with EMC storage and IBM AIX for some time in the past and I remember that led 554 on boot at one time was very likely. We got a fix (I will have to check my e-mail and see whether we fixed it ourselves, got a fix from IBM or from EMC). We also noticed that we often could not create an mksysb (since AIX couldn't determine the disk it booted from) - we fixed that by replacing 'pprootdev' (a Powerpath script from EMC) with our own version.
I will check my e-mail and get back here with some recommendations. Also, some of the suggestions here, sound pretty good to me.
------------------------------
Richard Westerik
Principal specialist
Simac IT NL bv
Ede
+31651575123
Original Message:
Sent: Fri April 14, 2023 01:45 PM
From: Luis Alberto Rojas Kramer
Subject: Error 554 with DELL (EMC) SAN Storage boot disk.
Hi experts..
We have an issue here at AIX booting time with EMC (Dell) SAN Storage System due path lost.
After seeing Kernel Starting message, suddenly stops at 554 LED code because booting process can't find the path from it had been accessing the IPL disk.
EMC told us we should reduce to just one path only from Storage to Server to avoid use other path which the Storage System is not offer the active path to these boot disk.
But doing that, is a problem and represent a task which takes SysAdmin time and could incur in risks by human errors every time an LPAR needs to be booted; like unmap virtual FibreChannel adapters or unconfigure SAN Switches zoning or delete WWPN from the host attachment config at Storage System.
Does somebody know how to avoid this problem happen?
We thought that maybe a vscsi disk from VIOS to LPARs could help, but, I can't believe every customer with EMC Systems should live with that problem.
Regards
------------------------------
Luis Alberto Rojas Kramer
------------------------------