A new behaviour when creating fully allocated volumes on SVC/Storwize V7.5 and later

ORIGINALLY PUBLISHED 14th October 2015

The support team are telling me that a lot of people are surprised by a new feature we introduced in SVC 7.5, which we didn’t do the best job in describing or documenting.  What you are seeing is a new Running Task appearing in the GUI when you create a new fully allocated volume that says “Volume Format” – a lot like this:

image

In the Volumes view – the volume is also shown as “Online (formatting) “

image

So what’s going on? Let me tell you.

The problem we are solving is that when you are using Fully allocated volumes in SVC or Storwize, we used to just give the volume to the host with whatever data happens to be stored on the managed disk.  Every now and then, that causes a problem because the managed disk contains data from other servers, and that can be a security risk.  Even more occasionally the start of the volume has Filesystem metadata on it which the operating system detects causing lots of confusion from the server teams (“Why does my new volume already have a file-system on it?”).  Thin provisioned volumes and compressed volumes don’t have this issue because of the way these functions work – they are always full of zeros until the host writes to it.

So in 7.5 we introduced a new feature which fills all fully allocated volumes with zeros as a background task whilst the volume is online – and we enabled it by default.     This means that any type of volume (fully allocated, thin provisioned or compressed) will all be filled with zeros when the host try to read them.  (We’ve actually had a feature which does this since SVC 1.1 – but the old version of the tool took ages to complete and kept the volume offline whilst it was filling the disk with zeros.)

If you click on the Volume Format running task, you will be able to see the estimated time until it completes.  For my quick test I created a 100GB volume and it was predicting 13 hours for completion – so your real volumes which are 2 TB or larger will probably say they are going to take weeks to complete.

My simple advice is just to ignore this.  It doesn’t affect the ability of the host to use the volume – so there’s no real reason to worry about it.

If you really want it to finish faster then you can modify the “Mirror Sync Rate” of the volume (click properties on the volume then press edit) to a higher number and the system will initialize the volume more rapidly.  However you may want to remember to reduce it again when it’s finished so you don’t get any nasty surprises next time you use Volume Mirroring to perform a data migration of that volume.

Hope this helps allay you fears about this new behaviour – we did document this in the fully allocated volumes section of the knowledge center – but we called it “Quick initialization” there – so no amount of searching would have helped you find this detail.

A last note – if you really don’t want this behaviour – then you can disable it in the advanced section of the create volumes dialogue box, or with the -nofmtdisk flag on the command line.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: