Wednesday, November 28, 2007

3ware thinks RAID-6 minimum drive count is 5

This is simply a warning for others. I picked up a 3ware 9650SE-8ML RAID card two days ago, and 4 drives to go with it. I figured I'd go RAID-6, 2 drives worth of date and 2 drives worth of parity..... wrong! Apparently 3ware's algorithm requires a minium of 5 drives. Wowsers. RAID-6 should be a minimum of 4. Now in fairness, their USERGUIDE for the card states the minimum in the section on adding raid arrays, but their datasheet, nor their references RAID-6 tech brief document say this. In fact, their own RAID-6 tech brief says RAID-6's minimum requirement is 4 disks. Oh well.


Friday, November 23, 2007

Down the drain goes the dollar...

For those who plan on traveling, the Dollar to Euro is now 1.4924 .
And the dollar fell to ¥107.68 Japanese yen.

Yikes. This sucks. Read it all.

CPU power usage

Core 2 Extreme QX9650 - 3.78W @ idle / 72.98W @ load ( $1,149.99 )
Athlon 64 X2 6000+ - 11.79 watts @ idle / 119.47W @ load
Core 2 Duo E6300 B2 - 12.02/45.49
Core 2 Duo E6300 L2 - 8.62/37.36
Core 2 Duo E6700 - 15.78/66

Time for RAID-6

Areca ARC-1220 PCI-Express 8 Port SATA RAID Controller
$479 -
$519 -

$129 - Areca ARC-6120 Battery Backup Module SC-ARC-6120
$299 - Lian-Li PC-201B
$90 - CORSAIR XMS2 2GB (2 x 1GB) (plus $40 rebate)
240-Pin DDR2 SDRAM DDR2 800 (PC2 6400)
Dual Channel Kit Desktop Memory Model TWIN2X2048-6400C4

$109 - AMD Athlon X2 BE-2400 Brisbane 2.3GHz 2 x 512KB L2 Cache Socket AM2 Processor
$88 - SeaSonic S12 II SS-380GB ATX12V 380W Power Supply

Seems the WD Enterprise drives have 5-year warranty, like Seagate, and given my trouble with Seagate drives lately, I'll probably switch back to WD for the power savings (GreenPower line).

WD RE2-GP 750 GB - WD7500AYPS

Karting in SoCal


Sending SMS via email

Virgin Mobile:

where phonenumber = your 10 digit phone number

Building the ultimate Linux-based music server

Get some green back from Seagate

Lucky for me I still have all my receipts.

The Pursuit of Streaming Happiness

Streaming to iPod touch or iPhone


Streaming MythTV from MythWeb using Flash

These LCD enclosures look interesting..

Thursday, November 22, 2007

I seem to recall some similar symantics worked for Clinton...

Frankly, I couldn't care less what Clinton did, but I do find a lot of these patents to be a crock of sh**.

The USPTO is quite frankly not working for the people, and instead is working as a tax collection arm of the US Government.

Free radio from Sirius w/ 49.99 pre-paid subscription

Not bad. I like the Area 33 line up better on Sirius than XM, and it looks like the merge *should* go through, yielding hopefully the best of both worlds. With a free radio in this deal, it doesn't hurt to start out with Sirius.

Glofiish X800 and M800
109 x 55 x 16 mm
4.3 x 2.2 x 0.6 inches

eTen glofiish M800
4.5 x 2.3 x 0.7 inches (114 x 58 x 18.4 mm) nearly same size..
GSM Quad-band phone capable of global roaming ( 850/900/1800/1900 MHz)
UMTS tri-band 3G phone capable of global roaming ( 850/1900/2100 MHz)
*awesome, HSDPA
VGA screen -- w00t w00t
slide-out keyboard
6.3 oz (178 g) -- little hefty

Hrmmm... looks like I may have to switch back to (gasp) Windows Mobile OS again.

Damn it.

No Sony Ericsson W960i for me :-(

I currently own a W800i which all-in-all I have loved, and yes I paid a lot for it as I live in the US where no carrier released the phone to market. I like it better in some respects than the W810i that was released and carried by several US Cellular providers. So here I read about an upcoming model, the W960i, that looked like a perfect replacement for my phone:

Alas, the problems are the networks!?!
  • UMTS 2100
  • GSM 900
  • GSM 1800
  • GSM 1900
Doh! So what's the problem you ask? Well, here are my requirements:

1) Good interface. Check!
2) Usable Media Player. Check!
3) GSM (for Intl Travel). Check!
4) 3G or higher. Check....sorta. Here's where the problem is.

You see, in the US, we're more or less using a standard losely named 3.5 G, HSDPA. It's superior in some ways, but works in the 1800 and 1900 MHZ frequencies. This phone only does 3G (UMTS) at 2100. Argh!

So next someone might as, why not K850i ? One word: OS. The OS on the W960i is the vernerable Symbian OS, which means I get to use the Opera browser! K850i simply uses the SE developed OS which doesn't lend itself to developing any apps for it.

Oh well. Damn you Sony Ericsson for making me jealous for something I really can't use.

Alas, it doesn't seem to be gaining any traction in the US (networks maybe??):

Oh, and damn you Apple for releasing the iPhone with dull-EDGE. I won't consider buying until it has HSDPA. I just won't buy phones like George Lucas wants people to buy multiple versions of Star Wars.

Wednesday, November 21, 2007

SmartStor NS4300N is useless and aweful

And Promise tech support is worse. Why? Let's talk about this total piece of garbage.

I bought this thing around a month ago. I populated it with three 750 GB Seagate NS series drives. First thing I notice is this thing isn't quiet enough. At 30 feet, I can still hear the fan. Already I'm unhappy with it. Installed the software to access the unit (I hate installing software to gain access to a WEB gui!!). Also, I needed to change my IP address to a 192.168.0.x address (as expected) to reach the unit. Opened the web interface, reconfigured network, configured device as RAID-5, and rebooted the unit. First notice, web interface is fine, but lacks detailed information for the system status. That point would become more troublesome later. Setup shares for Windows and NFS, and start access the device. Initial thoughts on usage is that store rates are rather slow. After several attempts, speeds drastically increase, but then drop back down again. Why? I'll never know.

Next came the start of my real troubles: hard drive bad sectors.

One of the Seagate drives started exhibiting troubles, but yet the RAID never failed the drive. Not good. The unit stated everything was healthy and functional. Great. However, the so-called "eventlog" would *sometimes* be littered with the following:

Nov 5 11:58:39 WARNING BSL log disk 2 at LBA 0x0564f4fb1 cleared
Nov 5 11:58:39 WARNING BSL update on disk 2 at LBA 0564f4fb1
Nov 5 11:58:39 WARNING Task 20 disk error on disk 2 at LBA 0x0564f4fb1 (Length 0xa) with status 51
Nov 5 11:58:29 WARNING BSL log disk 2 at LBA 0x0564f4fad cleared Nov 5 11:58:29 WARNING BSL update on disk 2 at LBA 0564f4fad

Doesn't look very healthy to me. Neither did the sound of the drive during these attempted writes. R/W performance also halted basically during those events.

At this point, I'm already feeling completely unsafe with this product, but boy was I over-estimating this thing as you'll soon see.

I add another 750 GB that I thoroughly tested before hand for 48 hours using a linux tool called 'badblocks'. Next I pulled the bad one, and the RAID rebuilt fine after 11 hours, as shown below:

Nov 9 08:55:30 INFO RAID status: "FUNCTIONAL". The NS4300N (thegrid) volume "/VOLUME1" is functioning correctly.
Nov 9 08:55:29 WARNING Rebuild on array 1 completed
Nov 9 07:48:18 WARNING Rebuild on array 1 90%
Nov 9 06:41:02 WARNING Rebuild on array 1 80%
Nov 9 05:33:40 WARNING Rebuild on array 1 70%
Nov 9 04:26:24 WARNING Rebuild on array 1 60%
Nov 9 03:19:03 WARNING Rebuild on array 1 50%
Nov 9 02:11:42 WARNING Rebuild on array 1 40%
Nov 9 01:04:26 WARNING Rebuild on array 1 30%
Nov 8 23:42:21 WARNING Rebuild on array 1 20%
Nov 8 22:34:38 WARNING Rebuild on array 1 10%
Nov 8 21:27:00 WARNING Rebuild on array 1 started
Nov 8 21:27:00 WARNING Disk 2 unplugged
Nov 8 21:26:56 WARNING RAID status: "CRITICAL". The NS4300N (thegrid) volume "/VOLUME1" is not functioning correctly.

No more troubles right? Wrong. Several days go by, continuing to move data from older non-raided drives to this unit at a blazing speed of around 5 MB/sec (ha), when all of a sudden the filesystem goes read-only, and won't go back to read-write without a reboot. I could repeat this scenario several times throughout a day. What the heck is wrong with this thing? "Eventlog" say nothing:

Nov 17 19:31:14 INFO System is starting to work.
Nov 17 19:29:11 INFO System is rebooting.
Nov 17 19:04:57 INFO System is starting to work.
Nov 17 19:03:03 INFO System is rebooting.
Nov 17 18:53:39 INFO System is starting to work.
Nov 17 18:51:27 INFO System is rebooting.
Nov 17 18:10:45 INFO System is starting to work.
Nov 17 18:08:46 INFO System is rebooting.

Well, that's not good of the device to not report a problem. Darn it, this thing sucks. I want it fixed, this is a poor quality product. So next I log a case with Promise Support on the web. I get an email back with a case number from an automated system. The case details I put were as follows:

My NS4300N has been giving me trouble lately. I purchased it 16 days ago, running three Seagate 750 GB drives in RAID-5 configuration. One of the 3 drives had bad sectors and was not being failed by the NAS, but I replaced it with a known new good drive (tested with linux util badblocks for over 48 hours), replaced over a week ago. Now I am regularly having problems with the mounted filesystems going Read-Only, however the event log does not show anything. When this happens, my only solution is to reboot the unit, which appears to trigger on restart a filesystem check (longer boot time than a reboot when working fine). I suspect that there are filesystem troubles, but there is no way I can test that or see that. Event log only shows:
.... same as above.....
As you can see from the timestamps, I can pretty readily fail this device.
Please advise.

The next day, I call to check on the status. I am told an engineer will call me back. An hour later one does, I explain what's going on, and he tells me to backup my data and reformat this sucker. Well isn't that just great. I say no, I want to find the problem before I have to do that extreme measure. Frankly, if this can't be fixed any other way, I won't be using this product. He says he can't get into the linux kernel level, and I say find someone who can. The net result is that he closed the case on the me the next day with the following solution:

Case Solution:
Hi, Try moving all the data off the unit and deleting and recreating the array. It might be compromised because of the bad disk. The integrety of the array might be failing. Move the data and delete the array and recreate it so it creates the nfs filesystem over.

Completely unacceptable. Thanks Promise support for closing my case without my ok! Thanks Promise for not investigate my problem! Thanks Promise for providing a useless solution to a problem that may, and probably will, occur again taking my data with it!

With some deduction, and a little help from the web and an exploit, I can pull the "real" dmesg output from this device, something Promise support could not/would not do :

eth0: Link is up
eth0: Flow control is on
EXT3-fs error (device dm-0): ext3_new_block: Allocating block in system zone - block = 247889925
Aborting journal on device dm-0.
EXT3-fs error (device dm-0) in ext3_prepare_write: Journal has aborted
ext3_abort called.
EXT3-fs error (device dm-0): ext3_journal_start_sb: Detected aborted journal
Remounting filesystem read-only
EXT3-fs error (device dm-0) in start_transaction: Journal has aborted
__journal_remove_journal_head: freeing b_committed_data
__journal_remove_journal_head: freeing b_committed_data

This isn't good at all. Either the filesystem has corruption (maybe in the superblocks) or the kernel has bugs. For the former, shouldn't the RAID-5 have protected against this??? Yeah, one would think. Great product, real great product.

Needless-to-say, I'm tossing this lemon. Picking up a RocketRAID for a little more than this thing cost, and running RAID-6 on my own linux, where I'll get useful things like disk-scrubbing, real event logging, and S.M.A.R.T. data, you know, things to actually PROTECT your data.