CoinFX
searchregister login

 

 

 

 

 

Home

PostHeaderIcon OpenFiler - Growing an XFS Volume

I've done some digging around the OpenFiler forums looking for any tips on growing a large XFS volume with out  loosing data.

I've got an areca ARC-1260 controller that allows for Online Capacity Expansion, and after adding a few new 1.5TB drives I wanted to grow the XFS volume in OpenFiler to see my new space.

After a bit of digging I did find an article, where a walk-thru was posted. Here are the steps below, trust me I was nervous too... as you must remove and recreate the partition the data sits on.

I tested this with an OpenFiler VM for a comfort level around doing it to my production data. On both systems I had 0 data loss.

Original thread - http://forums.openfiler.com/viewtopic.php?id=1612


The steps are as follow : (Some screen caps at the bottom of this article.) This can be done via direct console to OpenFiler or via SSH into OpenFiler

In step 1a (print) I never did get an Error about the backup GPT table not being at the end of the disk. If you do not either, just continue on thru the steps

Be SURE to make note of your disk geometry, start (typically 0.017) and end.


1.  parted /dev/ (e.g. /dev/sdb)
1a. print

make note of the last number from Disk Geometry

"Error: The backup GPT table is not at the end of the disk, as it should be.
This might mean that another operating system believes the disk is smaller.
Fix, by moving the backup to the end (and removing the old backup)?
Fix/Cancel?" fix

1b.     mklabel gpt
1c.     print - it will show something like:

Disk geometry for /dev/sdb: 0.000-5722045.500 megabytes
Disk label type: gpt
Minor    Start       End     Filesystem  Name                  Flags

make note of the last number from Disk Geometry

1d.     mkpart primary 0.017 5722045.500 (fill in your start  number as 1a. print states and end as the last number from Disk Geometry)
1e.     quit


2. stop all services that use the partition (smb, nfs, etc); ldap is not necessary to be stopped, this can be done in the OpenFiler GUI
3. umount /mnt//
4. vgchange -an (disable all volume groups)
5. partprobe (reread partition tables on all disks)
6. pvresize /dev/ (resize the physical volume eg: pvresize /dev/sdb1)
7. vgchange -ay (enable all volume groups)
8. mount -a (remount all disk partitions)
9. start all previously running services
10. xfs_growfs /mnt//
11. Go to volumes in the web administrator and edit properties of the volume to expand the size


The disk I wanted to grow was /dev/sdb

My original partition was  - Start 0.017 End 4291526.854

And the new partition size was going to be - Start 0.017 End 5722045.000


Parted OpenFiler

As you see, the disk geometry is 0.000-5722045.00 yet the partition is only 0.017-4291526.854

Upon completing step 1d and running print you can see my partition is now sized correctly at 0.017-5722045.00

Lastly we have to grow the physical volume, Keep in mind the Disk itself is /dev/sdb but now we are working with the physical volume, thus using /dev/sdb1


Last steps to be completed from console are growing the XFS file system. The mount point I'm using for /dev/sdb1 is /mnt/san/media

This will allow you to grow the volume inside the OpenFiler GUI (show below)


As you can see, my volume now has 25% more free space, growing from 4190.94GB to 5587.91GB

This was completed with out loosing any data on my volume.


Please please be sure to make note of the disk geometry prior to doing any of this, the data typically starts at 0.017, do not start the partition at 0.000.

I am not responsible for any data loss by following these steps, please be sure to have backups and test before performing this against any production data volumes.

Last Updated ( Thursday, 22 July 2010 22:31 )

 

PostHeaderIcon New gear in the rack.

It's been a long weekend, with more down time than I expected....

 

But, it's almost ALL done (atleast for now).

I've added new hardware, re-racked my server rack, re-wired the whole thing as well. I'm now in the process of usin vSphere 4 Storage Migration to move all my VMs off my Primary file servers SATA disk onto my new NFS server.

 

Whats new, whats changed?

 

I've added;

 

Linksys 24port 10/100/1000 PoE SRW2024P Switch

2x HP Proliant DL360 G5 servers

1x Dell PowerEdge 1800

1x IBM XSeries 306m

 

 

The 2 HP 360s are now my vSphere 4 ESX hosts, these each have 2 Quad Core Xeon chips at 3GHZ with 8GB of RAM (I need to upgrade RAM and NICs).

The Dell PowerEde 1800 is going to become my Netbackup 7.0 Enterprise server. (dual proc 3.2GHZ with 2GB of RAM

And the IBM 306m (dual core 3.0GHZ, 2GB of RAM) is now hooked up to my PowerVault 220s. This puppy also has 4x Broadcom NetXtreme 1GB NICs teamed together. This will be the new NFS server for the vSphere farm.

 

 

More details to follow (as well as pics, check the gallery)

Last Updated ( Monday, 29 March 2010 21:28 )

 

PostHeaderIcon The dust settles...

Or has it??

 

Well, as stated prior, I have all my data migrated over to the storage on the new areca ARC1260 controller... I've also completed an on-line RAID level migration from RAID 5 to RAID 6, then completed the first of many on-line capacity expansions, adding an additional 1.5TB drive to the array.

This will soon be followed up with 2 new (RMA'd) 1.5TB drives to grow the entire volume to a 8 drive x 1.5TB RAID 6 array.

Sadly the PSU I had ordered,hec Zypher... Failed. I RMA'd this device and it appears NewEgg no longer carries this PSU. Thus I must pic out a new one. I also orderd a Coolmaster PSU for my new Mediacenter, sadly this was DOA.

Other occurances this past weekend?

 

I installed my first (of many...or atleast 1-2 more) HP DL 360 G5's

 

This 1U puppy has 2 Procs, Quad core, Xeon 3GHZ, sadly only 4GB of RAM (trust me I'll upgrade this.)

These will replace my DL740G1 for my VMWare farm, these support ESX 4.0 =)

 

Lastly.... My NetGear 1 GB switch died on me... so now all my network took a trip back to the 100MB days...Hot Tub Time Machine??

I plan on replacing this with a beefy Cisco or Linksys (one in the same these days eh? lol) switch for some fun!

And I guess I added 4 320GB SATA disks to a RAID 5 array on my areca. This will be used for NFS Storage for VMWare as well as VDR backup space!

 

So what next?

 

RMA the Coolmaster DOA PSU

Purchase a new 1000W+ PSU for the file server

Purchase new 1GB switch

Figure out VMWare upgrade path to get to ESX 4.0 (on 360G5's??)

Build NEW mediacenter

 

..... the list goes on.

 

 

PostHeaderIcon Data migration

I've added my new 4x 1.5TB drives, set up a RAID 5 pack,  and begun the data migration off the old data stripe.

Estimated migration time is about 33 hours.

 

I'm moving data off an existing 4 drive stripe (Samsung Spinpoint EchoGreen 1.5TB drives), and platering the data into a new 4 drive RAID 5 volume (Same Sammy drives).

I plan on doing a RAID level migration to RAID 6 once the hard drives are freed up and tested (I had a few hiccups I think on possibly 2 of those drives). Thus I'll have my original wish of an 8 drive RAID 6 volume.

Thank you areca ARC-1260!

To get there I'll first have to free up the stripe volume of 4 1.5TB drives. once I have all the data moved over to my new RAID 5 volume, I can then stress test the old stripe drives.

Once I validate they golden, I'll add 1 to the existing RAID 5 volume, the ARC-1260 supports RAID level migration, allowing me to add the new 5th drive and convert the array from RAID 5 to RAID 6.

Once the data rebuild completes, (yes this is all online, with out losing data) I can then add the remaining 1.5TB drives to the volume with OCE.

OCE is Online Capacity Expansion, another goodie brought to you by the ARC 1260 card. Yeah... it was over $800 but WELL worth it, I've actually debated upgrading the RAM onboard from 256MB to the max 1GB, and adding the battery backup option.

 

Last Updated ( Thursday, 04 March 2010 20:17 )

 

PostHeaderIcon Data migration complete.

Well the migration of my data off the 4x 1.5TB stripe onto the 4x 1.5TB RAID 5 is complete. Sadly I had to do this twice as I hit the wrong button and wiped out the copy once it was complete...

 

I did end up losing maybe 20GB worth of data due to bad sectors. If I had a choice I would send back ALL 8 Sammy Spinpoint Echo Green 1.5TB drives and get Seagates or WD's or anything else. These drives are junk! I've replaced over 50% of the drives thru RMA... I hear horror stories about Samsungs Warrenty returns, so I'm hoping I don't have to deal with it.

 

The system is now performing an online RAID level migration (thanks to the areca ARC-1260) from RAID level 5 to RAID 6 with a newly added 1.5TB disk. Once complete, I can add another 1.5TB disk and grow the volume to include the additional 1.5TB drives I have on hand.

 

PostHeaderIcon WinCache in action

Wincache is a PHP cache module for Windows and IIS 7, I've had it installed for awhile, and I guess I'm decently happy with it. I still need to trim down my image sizes, as this is all hosted on a residental cable connection.

 

Heres how Wincache performed in a 30minute window.

 

 

Last Updated ( Tuesday, 16 March 2010 15:50 )

 

<< Start < Prev 1 2 3 4 5 6 7 8 9 10 Next > End >>

Page 4 of 10