Mark Post wrote:
That's is all that's necessary.  And, again, if you make sure that when DASD 
volumes are added, that the proper configuration files are there, it should 
work just fine.  However, consider the case where / is on an LV, and you just 
added another volume to the VG.  That's big time trouble.  So, to try to be on 
the safe side, the SUSE folks have apparently decided to err on the side of 
caution.  Any active DASD volumes at the time mkinitrd is run get added to the 
initrd.  If you add DASD via YaST, that all gets done for you, and you don't 
have to worry about it.  If you do it manually, you better know what you're 
doing, either in the setup, or the recovery effort.  :)


Wait !

I'm not denying that actually going through a mkinitrd/zipl cycle is a
bad thing (I might have said something in the liking.. but the pros
outweigh the cons here)! As you said, adding a PV in the VG in which the
LV hosting the root fs is is a very valid candidate to go through the
cycle (I know nothing about EVMS, but I guess the issue would be the
same) - based on the fact that you don't want to probe for each & every
DASD in the config - as indicated by various people - yet I am still
wondering why - but that's another story[1]...

So, doing a mkinitrd/zipl whenever a DASD is added just 'in case' (it
has influence over the root fs) - seems like a sane precautionary measure !

What I was (and still) curious about, is the (eventual) requirement to
mkinitrd/zipl for a volume that not whatsoever related to the root fs !

But then again.. maybe there isn't any such requirement ! In which case,
then.. My question was moot to start with[2] - and I'll go back to my
corner !

--Ivan

[1] If you were to probe every DASD in the config, the VG would get it's
PV fulfillment, and there would be no need to mkinitrd/zipl.. but then
we're back to the 'too many volumes take too long problem !'.

[2] But then again, that's why we love the Open Source/Free Software
(take your pick) environment[3].. We *KNOW* what's going on - so -
instead of bickering about unknown quantities, we're arguing about
concepts because the implementation is public... We're not arguing about
whether byte 2 bit 1 should be 1 or 0 for something to work.. but
instead, we're arguing about the meaning of byte 2 bit 1 !

[3] Then again, if the whole discussion is annoying people, it's
possible there are better venues for this, although I don't think the
linux-s390 kernel ML would be a better choice !

----------------------------------------------------------------------
For LINUX-390 subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO LINUX-390 or visit
http://www.marist.edu/htbin/wlvindex?LINUX-390

Attachment: smime.p7s
Description: S/MIME Cryptographic Signature

Reply via email to