Creating a full backup from within a damaged FreePBX VM,

[Retiaquester]

So I have seen some options in the forum but am asking for advice.

A little background. Our server went down earlier this week. And since there is
a cold spare I assumed it would be a quick fix. (I was in for a treat).
I removed 2 drives configured as a 1TB Raid 1 VHD from server A and put them in B.
Nearly identical hardware, nearly the same raid controller. Nearly the same age.

• However the Raid was recognized as Foreign
• Needed Parity-check.
• Needed in the process rebuilding
• and after 8 hours and 65% 1 drive got out of Sync, and was damaged.

So after lot’s of work I got the Datastore to mount and the VM’s to show up.
Both the the Windows Server and FreePBD booted fine and seemed OK.
But the two virtual machines took damage. When I copy the VMDK’s it throws a input/output error.

Today I salvaged the Windows server by making a image from within the virtual machine with,
a live booted Macrium Reflect. After extensive check’s the databases were compared to a four day old backup, and everything seems unharmed.

I would love a similar approuch for the FreePBX VM. I did get it to make backups and got it out.
But the machine itself is quite a job because it’s customized with UseCallManager Patch.

One note, The Windows image took damage in 3 sectors. Since I could pinpoint the files, those were not important. But the image process failed with Windows Server Building Disk-Image tool.
Yes I need to work on my backup strategy.

Please let me know what you think, any idea’s.

[Retiaquester]

So I’ll answer my own topic.It might be usefull for others when presented with similar issues.

After many attempts to fix the filesystem from the Vmware server known as VMSF, it turned out to be unrecoverable. Even after using a command known as VOMA. (a Chkdsk, fsck) command for VMFS 5/6 partitions.

I ended up creating for every Virtual machine (including FreePBX) a extra VMDK (Virtual Disk) slightly larger than the partition. The new VMDK were stored on a known good Datastore, not on the damaged RAID.

Then I used a bootable ISO SystemRescue (formerly known as SystemRescueCd)

mounted both the VMDK with corrupt data and the empty VMDK
Imaged the FreePbx patition with ddrescue. ( 1 i/o error after 27 attemps by ddrescue)

Then used the original VMX (Virtual machine Configuration File) and recreated the VM. With a new primairy harddisk, and The second VMDK containing the rescued image.
Then reïmaged the primaire disk with it.

It turned out to have 4096 bytes damaged (4KB) and the Packages database file was corruped by this. After rebuilding this. A week back in use all seems fine without breaking the machine activation.

Lessons to take home. RAID 1 is not always transferable, from one RIAD controller to the other. Even if the controllers are the same Brand, and timewise from the same Period.