Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]


Groups > comp.sys.sun.admin > #173

Re: RAID1 volume maintenance

Subject Re: RAID1 volume maintenance
From Stefaan A Eeckels <hoendech@ecc.lu>
Date 2012-07-02 00:35 +0200
Message-ID <20120702003530.e7c95cf148fc1972ad51796b@ecc.lu> (permalink)
References <nospam.jmcourco-1783F1.07495325062012@news.supernews.com>
Organization E.C.C. sa - Computer Consultants
Newsgroups comp.sys.sun.admin

Show all headers | View raw


On Mon, 25 Jun 2012 07:49:53 -0500
John  Courcoul <nospam.jmcourco@removethis.me.com> wrote:

> So I was left with a SunFire V245 server that hosts several important 
> services with two internal 146GB HDDs in RAID1 mirror config,
> containing the OS and applications and an external SAN with the data
> files. One of the two internal HDDs has failed and I need to replace
> it with a spare I have procured and is currently in an antistatic bag.
> 
> As per the manual, the way to go about the replacement will be:
> 
> 1. Put the new HDD in the chassis and have the OS recognize it.
> 2. Partition the new HDD identical to the failed unit.
> 3. Add the slices on the new as submirrors.
> 4. Detach the submirrors of the failed HDD.
> 5. Unmount any non-mirrored slices on the failed HDD.
> 6. Remove the dead HDD and use as a doorstop, paperweight, etc.

That's not quite how I would do it. This procedure creates a new set
of submirrors, which is rather complex, error-prone, and quite
unnecessary. 

> Am I missing anything?
> 
> What's the command used to add the new HDD to the system without
> having to do an INIT 6 and reboot the system?

Here's how I replace a failed SVM-only disk:

First, remove the dead disk, and plug the new disk in the bay you've
removed the dead disk from. The V245 supports hot-swapping.

Then, use the following command to copy the vtoc from the active disk:

# prtvtoc -s /dev/rdsk/c5t1d0s2 | fmthard -s - /dev/rdsk/c5t2d0s2

(assuming that the running disk is c5t1d0s2 and the new, empty disk is
c5t2d0s2).

Finally, use metareplace to sync the slices:

# metareplace -e themirror c5t2d0s0

etc.

> Do I need to format the slices and create UFS filesystems on them
> before attaching as submirrors, or do I just do the attach with
> METAREPLACE and let it synch itself?

No, fmthard takes care of partitioning, and metareplace only needs a
slice. 

Take care,

-- 
Stefaan

Back to comp.sys.sun.admin | Previous | NextPrevious in thread | Next in thread | Find similar


Thread

RAID1 volume maintenance John  Courcoul <nospam.jmcourco@removethis.me.com> - 2012-06-25 07:49 -0500
  Re: RAID1 volume maintenance Stefaan A Eeckels <hoendech@ecc.lu> - 2012-07-02 00:35 +0200
    Re: RAID1 volume maintenance John  Courcoul <nospam.jmcourco@removethis.me.com> - 2012-07-02 03:11 -0500

csiph-web