I don't know Rackspace's procedures nor your procedures; but I can offer you a clean explanation of not being able to restore the individual backups. I'm kinda familiar how other kinda-cloud provider does backups.

All the backups are encrypted with a transactional key. There are separate keys for each VM and for each incremental backup. The keys are stored only as long as the accounting record for the VM is open. Once the VM is deleted and summarized for the purpose of charging all transactional keys are deleted zeroized (I forgot to use the proper lingo).

The encryption key deletion zeroization is a security precaution against many inside attacks as well as against compromised backup media.

The above procedure is actually an important selling point. If you don't do that internaly for the 3rd party data then you should start doing this as soon as practical to lower your own liability.

http://en.wikipedia.org/wiki/Hardware-based_full_disk_encryption
http://www.hgst.com/internal-drives/self-encrypting-drives/

I would just add one little comment to the above excellent summary.

Bitcoin auditor paid in bitcoins would open itself to a conflict of interest charge.

So lets see who'll volunteer to become an auditor and just preemptively sue for the breach of fiduciary duty.

Well, I'm not trying to blame you personally, or anyone else here.

The main drawback of various mesh technologies is completely nontechnical. Currently it became fashionable to rephrase all old multi-level marketing blurbage using the new terminology of mesh communication. It is a new way to fleece new suckers.

For current bitcoin{d,-qt} definitely spinning drive will last longer. As it is right now bitcoin{d,-qt} is a worst-case load for the solid-state drives:

1) it writes a lot of data that is never read: the database/log.* files
2) it has bad write amplification behavior: writes too small chunks of data that aren't properly aligned.

So you'll need either a spinning plater or a high-end SSD. Better ask your wife than me.

I don't get the fixation on mesh systems here on this forum. Bitcoin is a broadcast system! It now uses peer to peer mesh to emulate a global broadcast because the currently publicly available routing infrastructure doesn't support multicast. So the peer to peer is more of a workaround.

By "doesn't support multicast" I mean "openly". Most of the current ISPs are in conflict of interest because they support multicast internally to provide TV over IP; so they are motivated to hush it down and disable at the customer-premises level.

Moon bounce is the word. Let them shoot the Moon down to disable the radio and laser links!

https://bitcointalk.org/index.php?topic=10241.msg539358#msg539358
https://bitcointalk.org/index.php?topic=46656.msg555899#msg555899
https://bitcointalk.org/index.php?topic=53371.msg636246#msg636246

Support the free trade on Earth by banking on the Moon!

The health is almost everywhere at 100%; 98% being the worst. If you really think that you haven't jostled the laptop itself, then continue to monitor the G-Sense counter every day or several times per day, until you'll identify the problem. Maybe its a window striking the table you keep the laptop on? Maybe you are slapping the lid shut sometimes?

You have 16 "Current Pending Sectors"; those are awaiting remapping/fixing.  This correlates with 15 "G-Sense" warnings: some vibration problem or unstable workspace.
 
Yeah, probably aggresive power saving. The Start-Stop count is at 100 with 94 power cycles.

I always did just that. But I never sent the whole computer back, only the disk.
I always did "smartctl -a", but if you want to save space you can use "smartctl -A", "smartctl -l selftest" and "smartctl -l error". In would also print the results from the manufacturer's diagnostic: either OEM like Dell, IBM, HP or the actual drive manufacturer if this was a retail drive. I also have a duplex printer that I could configure to print 4 pages per one sheet of paper, so I couldn't claim saving the trees.

"-A" is the most interesting to look at. "-l error" looks the scariest if you suspect bad behavior from the service department.

"-A" could also give you away. E.g. you'll have "G-Sense Error Rate" high then its a proof of a dropped laptop. Same things with "Power On Hours", "Power Cycles" and "Temperature Celcius" if you were fiddling with the computer.

This is a bad advice. It gives you false negatives: chkdsk only scans inside partitions, but doesn't scan partition headers and various boot and recovery files.

Also chkdsk will be fooled with extended retry algorithms that are used in modern drives (with the exception of the most expensive enterprise drives). It will be silent about all the marginal sectors that are barely readable after many seconds of trial reads. You especially want to know about those in advance.

smartmontools are the best friend of smart disk drive users.

If you like the laptop then keep it and use it. Repair the bad blocks, monitor for any new bad blocks and get the replacement drive under warranty if you get any more bad blocks without a reason, like jostling.

What is worth more: a spare hard drive lying on the shelf or your time wasted with no laptop?

Edit: According to the IT-man I know, the most common case of early laptop drive failure is when people shift the fresh laptop on a smooth desk. The clean rubber feet make the laptop jump a little and then visibly oscillate while being shifted. That doesn't happen once the rubber feet are somewhat dirtier and worn out.

Don't let yourself get discouraged by the naysayers here. Casascius is a monocoinist polycoinist and markm is an innocent bystander. Actually makomk is a victim of wrath of a gang of monocoinists. I apologise for the confusion between markm and makomk. I also apologise to casascius for misspeling his name and incorectly declaring his ideological allegiance.
 
The knowledge you gain will be yours for your lifetime.

The calculation isn't so easy. The blkNNNN.dat files are maintained with buffering but without obeying proper lock protocol.

You could probably work around this problem by creating file system snapshots (shadow copies in the Windows parlance: vssadmin.exe is the tool to look at).

You could use either heavily recursive or high-precision floating point proof of work. Or both. My old post about those:

https://bitcointalk.org/index.php?topic=44423.msg537010#msg537010

I'm not sure what you are exactly doing.

"smartmontools" works the same under Windows and Linux.

"badblocks" runs only under Linux, and since it could be destructive you need to run it while Linux is booted from CD or USB in recovery mode.

Only worry about bad sectors if they can't be repaired. If you are going to do a complete reinstall then run "/sbin/badblocks -wsv /dev/sda" to completely rewrite and remap all possible bad blocks.

This is especially true in case of laptops, where bad blocks could be the result of an accidental jostle.

Actually, you are probably quoting the average fragmentation of the whole disk.

If you had a tool that shows just the fragmentation of the Bitcoin files you would get astounded. They are nearly maximally fragmented.

To quickly defragment just the Bitcoin directory:

1) copy it to a backup disk
2) delete the original directory
3) copy back into the original place

If you really can't backup onto a separate disk do at least zip, delete, unzip.

I don't know what software you downloaded. You should've gotten smartmontools. Then simply run info & capabilities; short and long tests. Edit: and display all.



C:\Program Files (x86)\smartmontools\bin>smartctl.exe -i -c /dev/sda
C:\Program Files (x86)\smartmontools\bin>smartctl.exe -t short /dev/sda
C:\Program Files (x86)\smartmontools\bin>smartctl.exe -t long /dev/sda
C:\Program Files (x86)\smartmontools\bin>smartctl.exe -x /dev/sda


Edit: once you boot Linux you can also fix/remap the bad blocks automatically using "/sbin/badblocks -nsv /dev/sda". But do it only after you've made a backup. Trying to remap bad blocks on a disk that run out of spare blocks typically ends in an even worse failure.

Just an interesting tidbit I've found on the 2nd pass through the Knights Corner documentation:
248/4 = 62 not 50.

Why could that be?

1) A leftover from Knights Ferry?
2) Yield with all 62 cores enabled would be zero?
3) Something else?

Please share your guesses.

Dear eldentyrell!

Please don't turn yourself into another copy of Luke-Jr. There's no point to gain from antagonizing anyone here. We all know that you've call-ed the bluff on the PR web front. But the opponent had bigger bankroll, better poker skills and raise-d over your call. It is unsporstmanlike and ungentlemanly when you try to upturn the table after the game on that table turned against you. Now just do like everyone else: wait for the showdown to see if this was bluff or semi-bluff.

Anyway, you are a pure software vendor and you have several times already said that you aren't willing to work with hardware vendors on porting your software.

Why would the moderators of this forum need to listen to your lectures on how the hardware subforums should be organized? Just move to the mining software subforum if you really can't stand the heat here.

Thank you for your attention to this matter.

Man, the Digilent boards are about as custom as the other FPGA hardware sold here. Digilent wouldn't exist without continuous support from the marketing funds of XLNX, which acts as a rich daddy and organizes group purchases by the educational institutions.

That XUPV5 board that eldentyrell had linked to is even more custom than any other Digilent's product. They were given away for free by SUNW/JAVA in an attempt to position SPARC as an open industry standard. If not for the marketing budget of Sun Microsystems only a handfull would ever be sold. The price tags on them are just for show and tax deduction purposes.