Linux Storage - Removing the standby array from RDAC

category: solutionz · date: 2010-09-30 · updated: 2026-10-03 · author: LALA

Linux Storage Solution · Config document · referenced from RDAC problems and LUN removal

noteThe "wanted state" outputs carry the same Time and the same file dates as the "faulty state" outputs, which suggests they were written by editing the faulty listings rather than captured after the fix; the notes do not say. The fix itself ends with a reboot and two check commands whose output the notes do not keep. The array WWNs are made-up values of the right shape.

Problem 1 of my notes: mppsrv01 saw two disk arrays, the main DS5300-SITEA and the standby DS5300-SITEB, which is a mirror of the main one. The notes call that a problem above all when LVM is in use, because devlabel cannot tell the attached devices apart. This document holds the faulty state, the state that was wanted, and the command set that got there: turn off array discovery at boot, take the standby array out of the persistent mapping, remove its .wwn file, rebuild the initrd, reboot, check.

Item	Value
Run as	`root` on `mppsrv01`, working directory `/var/mpp` for the `ls -l`, `cat ./devicemapping` and `rm ./…wwn` commands (the notes use relative paths)
Host	Red Hat Enterprise Linux AS 4, kernel `2.6.9-89.0.9.ELsmp`, two QLogic HBAs
Arrays	`DS5300-SITEA` (ID 0, main, kept), `DS5300-SITEB` (ID 1, standby, removed)
Edits	`/etc/init.d/mpp` run levels (`chkconfig`), `/var/mpp/devicemapping`, `/var/mpp/600a0b800012345700000000aabbcc02.wwn` (removed), `/boot/mpp-2.6.9-89.0.9.ELsmp.img`
Taken	8 September 2010
Driver version	RDAC `09.03.0B05.0331`

The listing

The faulty state, as mppUtil -a showed it:

bash

$ mppUtil -a

output 12 lines

Hostname    = mppsrv01
Domainname  = (none)
Time        = GMT 09/08/2010 10:50:35

---------------------------------------------------------------
Info of Array Module's seen by this Host.
---------------------------------------------------------------
ID              WWN                      Type     Name
---------------------------------------------------------------
 0      600a0b800012345600000000aabbcc01 FC     DS5300-SITEA
 1      600a0b800012345700000000aabbcc02 FC     DS5300-SITEB
---------------------------------------------------------------

The faulty state in the persistent mapping:

bash

$ cat ./devicemapping

output 2 lines

0:DS5300-SITEA
1:DS5300-SITEB

The faulty state in /var/mpp:

bash

$ ls -l

output 3 lines

-rw-------  1 root root 361 Sep  7 11:57 600a0b800012345700000000aabbcc02.wwn
-rw-------  1 root root 361 Sep  7 11:57 600a0b800012345600000000aabbcc01.wwn
-rw-------  1 root root  15 Sep  8 13:26 devicemapping

The state we wanted, as mppUtil -a should show it:

bash

$ mppUtil -a

output 11 lines

Hostname    = mppsrv01
Domainname  = (none)
Time        = GMT 09/08/2010 10:50:35

---------------------------------------------------------------
Info of Array Module's seen by this Host.
---------------------------------------------------------------
ID              WWN                      Type     Name
---------------------------------------------------------------
 0      600a0b800012345600000000aabbcc01 FC     DS5300-SITEA
---------------------------------------------------------------

The state we wanted in the persistent mapping:

bash

$ cat ./devicemapping

output 1 line

0:DS5300-SITEA

The state we wanted in /var/mpp:

bash

$ ls -l

output 2 lines

-rw-------  1 root root 361 Sep  7 11:57 600a0b800012345600000000aabbcc01.wwn
-rw-------  1 root root  15 Sep  8 13:26 devicemapping

The commands

First, the attempts to discover new arrays at boot were turned off, and it was checked that mpp is off in every run level. The notes number these two lines [1] and [2]; the check has no output in the notes:

bash

$ chkconfig mpp off
$ chkconfig --list | grep mpp

Then the array DS5300-SITEB was removed from the persistent mapping:

bash

$ mppUpdate -d DS5300-SITEB

output 3 lines

The module DS5300-SITEB was removed from the MPP persistence mapping list (/var/mpp/devicemapping).
Detected 2 QLogic Host Adapter Port(s) on the system
Creating new MPP initrd image...

The WWN record of DS5300-SITEB was removed, the initrd was rebuilt once more, and the machine was restarted so that the settings take effect:

bash

$ rm ./600a0b800012345700000000aabbcc02.wwn
$ mppUpdate
$ init 6

After the boot, the check that the server sees only the one array it should. The notes keep no output for these two commands; the "wanted state" listings above stand in for it:

bash

$ mppUtil -a
$ cat ./devicemapping

Reading it

Step	Why, as far as the notes say
`chkconfig mpp off`	`/etc/init.d/mpp` "discovers devices at boot"; the notes' own list of commands adds that with more than one array it is worth turning it off so that a restart causes no unwanted trouble. Without this, the next boot would find the mirror again
`chkconfig --list`	the check that the service is off in every run level
`mppUpdate -d DS5300-SITEB`	takes the line `1:DS5300-SITEB` out of `devicemapping` and rebuilds the initrd, so the mapping in the initrd no longer knows the array
`rm ./…aabbcc02.wwn`	`mppUpdate -d` does not touch the `.wwn` file; it had to go by hand, otherwise the driver would still hold the array's path table
`mppUpdate`	the initrd is rebuilt again, now without the `.wwn` file; as I understand it, the `.wwn` files are packed into the initrd, which is why a second rebuild was needed after the `rm`
`init 6`	as I understand it the driver reads its persistent state only when its modules load, so a reboot is the way to apply it; the notes reboot without saying why
`mppUtil -a`, `cat ./devicemapping`	one array, one line

Two small things in the listings. The ls -l of the faulty state lists the …02.wwn file before the …01.wwn file, which is neither name order (plain ls -l sorts by name, and …56… comes before …57…) nor what the earlier ls -ltr in /var/mpp files printed; the lines were most likely rearranged by hand when the notes were written. And devicemapping is dated Sep 8 13:26 while the mppUtil -a above it says GMT 09/08/2010 10:50:35; the notes do not say in which order the two were taken.

What the notes do not hold: whether the standby array stayed zoned to the host on the SAN after this (the eight physical devices would then still be there, and the second problem in the same Article removes exactly those), how DS5300-SITEB came to be a mirror of DS5300-SITEA, and what happened to the LVM volume group that prompted the fix.

Checked against Red Hat Enterprise Linux 9 and 10

As built	Today
`chkconfig mpp off`, `/etc/init.d/mpp`	Red Hat's 9.7 release notes list it as a known issue that "The chkconfig package is not installed by default in RHEL 9" and point to `systemctl`; the package is still built (`chkconfig-1.24` for 9, `1.30` for 10)
`mppUpdate` rebuilding `mpp-2.6.9-89.0.9.ELsmp.img`, GRUB legacy `menu.lst`	GRUB 2 with boot loader specification entries, `grubby` and `grub2-mkconfig`; the initrd is built by `dracut`
`devlabel` as the tool that cannot tell the mirrored devices apart	`devlabel` is a Red Hat Enterprise Linux 3 and 4 era tool: the package `devlabel-0.48.03` ships in CentOS 3.9 and there is no `devlabel` package in CentOS 4.x or 5.x, so even on this host it may well have been a leftover habit rather than an installed tool. Device identity is udev's job today
RDAC with two arrays	Discontinued with SANtricity OS 11.25; as I see it, a mirrored array seen twice would still be a problem for LVM under DM-Multipath, and the fix would be on the SAN (zoning, LUN mapping) rather than in a driver's state files
Red Hat Enterprise Linux 4	Retired 29 February 2012, extended life-cycle support ended 31 March 2017