[GLLUG] Advice on expanding an existing RAID 1 installation
damion.yates at gmail.com
damion.yates at gmail.com
Sat May 24 21:15:29 UTC 2014
On Sat, 24 May 2014, John Winters wrote:
> The current system has 2 x 1TB drives in a RAID 1 configuration.
> These are running out of space.
> I intend to get 3 x 3TB drives and would like to transfer everything
> to those.
Sensible, 3TB are dirt cheap atm, I recently got several myself.
> The Linux RAID Wiki has a page on extending an existing RAID array,
> but says I should start by removing one of my existing drives.
> Presumably this isn't necessary if I have a spare physical
> slot/controller slot/power lead? Rather than their proposed approach
> of remove=>add=>remove=>add, is there any reason I can't do
I'm pretty sure I've done what you suggested. However I recall having
bus errors (my m/board was a tad on the cheap side) which confused linux
enough that it decided to accept nulls read from the newer disk as
correct data and slowly wiped my array. That may have been a different
time and different system though, stressing my cheap m/board bus with 3
disks would have been mad.
Note that with a disk taken out of the array, you have a good backup
should anything go wrong before the sync finishes.
> Second question - I'm undecided about whether to put the 3rd drive
> into the system to start with (thus three drives up to date all the
> time) or keep it on a shelf for when it's needed. Does anyone have
> strong views on which is better?
You mention this is a drive for backing up various systems, but this way
you still only have two copies. You're more immune to user silliness,
but it wouldn't hurt to have a 3rd backup. I'd keep the 3rd disk as an
rsync-to-able backup, attached to another system. If you're able to
take it away in case of fire, or maybe simply another room to decrease
odds of a fire wrecking both rooms/disks then it's even better.
I'm in a vaguely similar situation to you. I have a main low power
netbook I keep on all the time at the back of our junk room (I don't
know what I'd do if it ever stopped, I don't think I can physically
reach it anymore!) for a general screen(1) session to email, irc, IM
etc. Many phones for the family which needs photos rsynced from,
various laptops, and a PVR (MythTV) recording everything good that's
still on freeview.
All of these rsync/nfs to a 3TB NAS (~£100) which is on all the time and
surprisingly reliable for the stupidly low cost. It's not raid where
your system is, and I would be irritated if its drive went. However I
then also back that up every now and again to a 3TB usb drive (~£100).
I now have 3 copies of my data with minimal effort or cost.
I can lose two devices without losing the actual data, but would have
some annoyance if any of the devices were to fail.
If the data isn't that important (pvr of shows that will be repeated one
day), I can remove it from the master device knowing I retain an easily
accessible copy (the NAS is nfs mounted on mythtv to be played via the
same ui) which it still backed up.
In your case your RAID1 system is obviously slightly out of date if only
powered once a day, so I don't really understand why it needs raid for
resilience to its data. In fact you'd be better off with the 2nd drive
doing a weekly rsync from the 1st after systems are backed up to the 1st
on a daily basis. That way if the backup is copying some corrupt/wrong
version of whatever you've been working on, you know you can roll back a
few days and restore.
Anyway that's just my thought on the matter.
I've used Solaris disksuite for years and loved it, I've also used
Linux's raid with mdadm and decided it was perfectly acceptable. But
then I've also had disasters which were, frankly down to the
complication or having used RAID which wouldn't have been there
otherwise. You're adding extra layers of complexity for minimal gain.
It's especially bad when people are using RAID as their only copy of
data. This is stupid as that data is one rogue rm away from being lost
Of course the decades of servers I've been involved with, I'm more than
happy for raid, as the sheer no.of machines means disks can fail with
some frequency and it's less of a pain to get a disk swapped then run a
few commands remotely than to have to rebuild.
More information about the GLLUG