Hello,

On a RS6000 server I have two disks (hdisk0 and hdisk1) which are mirrored in a logical volume rootvg.
Yesterday one of these disks failed. Diagnosis, physical death.
It has therefore been replaced by a brand-new.
In terms of parameters, hdisk1 is now N# 1 in bootlist, everything is OK (hd5 is on, bootable ...)

At the first start, the boot sequence is hold, and the front LED display shows error code 0552.
A small turn in the documentation teaches me that this means ipl_varyon is not running ok.

On an equivalent server we try to repeat the config.
If hdisk1 is removed , hdisk0 boots, if hdisk0 who plays the role of dead HD is removed, hdisk1 does not boot.

After a nice series of tests of all kinds, here's what I tell myself ...
Does the hdisk0 (the one dead) may be a flag somewhere that allows a boot even if only 1 of 2 disks of the quorum is present while hdisk1 have this same flag disabled thereby excluding a boot rootvg if 1 of 2 disks is missing.
Yet as part of a mirroring, the discs are equivalent, there is not one that would be "master" more than the other, is it?
The variable MISSINGPV_VARYON is set to TRUE in /etc/environment, I think it would solve the problem but not.
I can not find anything, anywhere (smitty, /etc/environment ...) which proves this.

That tells something to someone ?