Re: rootvg not on hdisk0 but on hdisk4
From: Piotr (nobody_at_now.where)
Date: 11/26/03
- Next message: Mike: "magic string to HMC"
- Previous message: Matt: "Re: rexec problem from NTPCs"
- In reply to: Andreas Schulze: "Re: rootvg not on hdisk0 but on hdisk4"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 26 Nov 2003 20:05:51 +0100
"Andreas Schulze" <b79xan@gmx.de> a écrit dans le message de news:
bq1l92$k743@news-1.bank.dresdner.net...
> "Piotr" <nobody@now.where> schrieb im Newsbeitrag
> news:bq0l89$b55$1@news-reader5.wanadoo.fr...
> > Hi,
> >
> > I don't think he has only 2 disks... Or his AIX would be quite
strange...
> > Looks like he has 2 internal hdisks and 4 external hdisk in a 2104, for
> > example...
> >
> > In this case (whatever you should have instead of 2104) and if you can
> have
> > your server for about an hour, the best you could do is :
> >
> > 1) Run a mksysb of your system
> > 2) Shutdown your system
> > 3) Unplug external disk bay
> > 4) Power on your system in sms mode
> > 5) Restore your mksysb (your internal hdisk will be the only disks
viewed
> by
> > AIX, so that, they will be hdisk0 and hdisk1)
> > 6) During the reboot after the mksysb restore, plug your external disk
bay
> >
> > You should have hdisk 0 & 1 for rootvg and hdisk 2,3,4,5 for other
> vg(s)...
> >
> > It's a little bit furious, but so fast and so simple...
> >
> > Regards
> > Piotr
> >
> Hallo Piotr,
Hello,
>
> he could have only two disks if AIX was installed on disks4/5 while other
> (older) disks where still present that were removed later. I could imagine
> that hdisk4/5 replaced a rootvg of 4 smaller/slower/older disks hdisk0-3.
> That would be not strange but quite normal because some people like to
keep
> a old system while preparing a new one in parallel. This will leave them
an
> easy way for rollback as you just change bootlist to return to your
original
> system.
>
Ok, it's possible.
> Secondly the information that AIX will discover hdisk4 and hdisk5 as
hdisk0
> and hdisk1 because that are the only disk it will "see" is much likely to
be
> wrong. AFAIK as long as information of hdisk0,1... remains in the ODM (as
> defined hdisks) there will be no other disk be able to occupy these names.
> It does make no difference whether the disks are internal or external. In
> both cases a pvid is linked to a hdisk definition which need to be cleared
> if it shall be reused.
>
When you restore a mksysb, the ODM is rebuild, using the "new" hardware
configuration...
> Thirdly even if there are disks that are suitable in respect to size and
> speed your method using the mksysb will take much longer than using the
LVM.
> IF there are disks in the system Peter wants to use he could unmirror one
> disk of the rootvg, remove that disk from rootvg, extend rootvg by the
> (present) hdisk0 and then mirror again. Set BLv and bootlist. Then
unmirror
> the second disk (4 or 5 depending on which disk he chose first) from the
> rootvg, reduce rootvg by that disk include hdisk1 into rootvg and mirror
> again. Check Quorum, set BLV and Bootlist. All this can be done while the
> server is up and running. Now comes the donwtime of the LVM method: one
> single reboot. Compare this to the time the server will be offline during
> replaying an mksysb. The mksysb method will compare even worse if the
disks
> are of a size not suitable so that Peter might have to clone the system.
So
> I still would prefer the simple approach. There are other occasions when
the
> mksysb is the first choice. It is not the preferred method for swapping
> disks within a server however.
>
Of course, it works, but it's a little bit longer... I just said it was an
other way to do it, not the only...
Depending on the disk size, the unmirror/mirror operation is most of the
time longer than backup/restore with mksysb.
> As I said before: knowing the LVM is a great way to start liking AIX.
;-).
Yes, when you tasted AIX LVM, it's hard to play with other Unix...
>
> HTH,
> Andreas
>
Regards,
Piotr
- Next message: Mike: "magic string to HMC"
- Previous message: Matt: "Re: rexec problem from NTPCs"
- In reply to: Andreas Schulze: "Re: rootvg not on hdisk0 but on hdisk4"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
|
