Gator127 Hardware Issues

First and foremost our most profuse apologies to all of our new customers on the Gator127 server. We know it has been up and down and we've been working round the clock with our data center to resolve the issues and we have now found the problem.

As you may or may not be aware, all of our shared and various other servers use RAID (Redundant Array of Inexpensive Disks) in order to protect against catastrophic hardware failure with one of the drives. Basically whatever data the server writes to one drive is also written to a second drive in case one of the drives fails. In theory and in practice this is perfect for hardware failures however it also means that any file system corruption written to one of the drives is "mirrored" on the other.

We have found the problem to be that the RAID controller (the device that does the writing to both of the drives) is a 32 bit piece of hardware with a 32 bit bus. What this means is that it can process data in 4 byte chunks, and then the next series of 4 bytes is sent so on and so forth. Well, the operating systems for this server was sending 64 bit information to the 32 bit raid controller, that means the operating system was attempting to send 8 bytes at a time to a device that can only process 4. What occurred was that this "extra" data was buffered to be sent down the line later, but eventually this buffer filled. When this occurred blocks (how data is stored on the drive) were written out of order, some lost entirely, and a general mess of things was made. This resulted in file system corruption which is what brought the server down.

This technical explanation is all fine and dandy but I'm sure you all want to know in plain English what this means. Basically it means we need to wipe clean the server, reinstall the operating system (an "OS Reload") and then restore from a backup we have that is 24 hours old. This means that any new data within the last 24 hours will be lost. We profusely apologies for this situation and assure everyone that all admins are working double time to rectify this grievous mistake. We will keep everyone updated with our progress and if anyone has any questions or concerns please do not hesitate to contact us.

Again our apologies for the frustration and inconvenience this issue has caused. We expect to be up and running shortly, and we will relay information as we receive it.

Best Regards,
Robert Stone

[kmaw]

It's a nice explanation...

[Hbug70]

I couldn`t have done it better.