ESXi 4.1 missing datastore after HW crash - no VMFS Label

[hrabovsky]

Dear all

On my one HP server the diskarray and then 2 disks have been changed. Now HW looks ok, but in ESXi the datastore is missing.

I tried force mount, but I see under Devices the Disk ( Local VMwareDisk ) without VMFS Label. It's possible to recover it? Thanks.

[a_p_]

... diskarray and then 2 disks ...

Can you please provide some details? What happened, this sounds as if the RAID was destroyed with two repaced disks!?

What's the output of fdisk -lu from the host's command line?

André

[hrabovsky]

Hi Andre

I'm waiting for the local guy to get fdisk output because I have no low level access ( ILO ) 

BTW. I think that RAID5 with 8HDDs didn't survived 2 failed HDDs. But it's strange. I've got this detailed information.

The controller for the Tape device - not for HDDs was changed, after restart HDD no. 4 didn't started, new disk was ordered,

but in the evening the HDD no. 3 failed, VM reported no Datastore. Technician cames in the morning, turned the server off

and replaced HDD no. 4. Then he started the server and all HDD were green. There is still 1TB logical drive on the controller,

VM can see this device but no VMFS volumes. I tried to rescan, but without success.

Message was edited by: a.p. - removed email footer

[a_p_]

This doesn't sound really good, be prepared to recover from backup.

André

PS: I removed your personal data (email, phone, ...) from your previous email reply.

[continuum]

The only thing you can say for sure at this time is that the partition tables have been trashed - thats quite normal and happens frequently.
Before you have not tried to recreate the partition tables I would not give up your data.

[hrabovsky]

Hi continuum

Till now no action, what do you suggest?

[continuum]

what does fdisk report ?

[hrabovsky]

the disk doesn't contain a valid partition table

[hrabovsky]

Hi

I'm trying to recover partition ( http://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=103660... ), but I don't know what to enter for the first sector,

in this KB is 128, but I think in my case I should use the default value 63 :

~ # fdisk -l /dev/disks/mpx.vmhba2:C0:T0:L0

Disk /dev/disks/mpx.vmhba2:C0:T0:L0: 1027.4 GB, 1027474350080 bytes

255 heads, 63 sectors/track, 124916 cylinders

Units = cylinders of 16065 * 512 = 8225280 bytes

Disk /dev/disks/mpx.vmhba2:C0:T0:L0 doesn't contain a valid partition table

~ # fdisk -u /dev/disks/mpx.vmhba2:C0:T0:L0

Device contains neither a valid DOS partition table, nor Sun, SGI or OSF disklabel

Building a new DOS disklabel. Changes will remain in memory only,

until you decide to write them. After that the previous content

won't be recoverable.

The number of cylinders for this disk is set to 124916.

There is nothing wrong with that, but this is larger than 1024,

and could in certain setups cause problems with:

1) software that runs at boot time (e.g., old versions of LILO)

2) booting and partitioning software from other OSs

(e.g., DOS FDISK, OS/2 FDISK)

Warning: invalid flag 0x00,0x00 of partition table 4 will be corrected by w(rite)

Command (m for help): n

Command action

e extended

p primary partition (1-4)

p

Partition number (1-4): 1

First sector (63-2006785839, default 63):

S pozdravom/Mit freundlichen Grüßen/Best regards

Jozef Hrabovsky

Phone: +421 2 58 266-150 Mobile: +421 905 295019 Fax: +421 2 58 266-168 E-Mail: jozef.hrabovsky@basf.com

Postal Address: BASF Slovensko s r.o., Prievozska 2, 821 09 Bratislava, Slovakia

From: Ulli Hankeln <communities-emailer@vmware.com>

To: <jozef.hrabovsky@basf.com>

Date: 12. 02. 2013 18:32

Subject: New message: "ESXi 4.1 missing datastore after HW crash - no VMFS Label"

[a_p_]

... but I think in my case I should use the default value 63

Why do you think so? Has the VMFS partition been created using an older ESX(i) version?

Anyway you can try both 63 and/or 128. For more details see http://kb.vmware.com/kb/1002281

André

[continuum]

dd if=/vmfs/devices/disks/mpx.vmhba2:C0:T0:L0 of=/tmp/hrabovsky-30mb.dd bs=1M count=30   

run the command and attach hrabovsky-30mb.dd
please do not change the name

I just see that you already created a new partitiontable - it would have been better to check the correct starting point first

[hrabovsky]

ok, the partition has been recreated with the default value, now I'm going to mount datastore, but it seems not to be mounted - vmfs label is still

missing , I'm afraid, the wizard is going to delete my old partition and to create a new partition

[hrabovsky]

Hi

file created and attached.

[hrabovsky]

Hi

any suggestion?

S pozdravom/Mit freundlichen Grüßen/Best regards

Jozef Hrabovsky

Phone: +421 2 58 266-150 Mobile: +421 905 295019 Fax: +421 2 58 266-168 E-Mail: jozef.hrabovsky@basf.com

Postal Address: BASF Slovensko s r.o., Prievozska 2, 821 09 Bratislava, Slovakia

From: Ulli Hankeln <communities-emailer@vmware.com>

To: <jozef.hrabovsky@basf.com>

Date: 14. 02. 2013 13:22

Subject: New message: "ESXi 4.1 missing datastore after HW crash - no VMFS Label"

[continuum]

you should recreate a partitiontable that uses a start offset of 128

that matches the start of the vmfs partition
but it looks like the beginning of the partition was at least partly overwritten with data that should not be there

I guess you will not be done when the partition table is fixed

please examine the next few mbs - does it go on with this pattern ?
”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9Í”9

[hrabovsky]

Hi

yes, the start offset of 128 was probably the correct one, datastore is back, but with incorrect capacity and without data

[continuum]

try to read the data with vmfs-tools from Linux

check your PMs - I give you a downloadlink for a LiveCD with vmfs-tools

[hrabovsky]

thanks all, no more wasting time, data are in the RAID probably corrupted, reinstallation and restore from backup started.