Personal computing discussed

Moderators: renee, morphine, Steel

 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Samsung 830 Corruption Woes...

Thu Mar 07, 2013 2:34 am

So I recently changed from a 3570k to a AMD 8350 for a few different reasons including it being free and something I was interested in doing. Anyway, I back up my PC on a fairly regular basis with Acronis. I have a 3xr0 array of 128GB Samsung 830s. I backed up my PC before switching out systems, which was actually rather painless. I got everything up and running, but started running into some weird problems. Whenever I try to make a new image with Acronis my computer BSoDs (something about a system process terminating unexpectedly). I thought this was rather odd so I did a bit more in depth troubleshooting. I ran chkdsk which popped up with some errors and fixed them, then run chkdsk /r which results in a BSoD on test 4. The resulting BSoD was "Bad system config info". Looking this up shows a lot of people with bad hard drives or in a case where windows is completely inoperable (basically the whole system is corrupt).

Running sfc /scannow also resulted in some errors which were fixed. The weird thing is when adaptive read ahead is off (but caching is on) on the controller for the raid array chkdsk and sfc don't turn up errors. chkdsk completes perfectly fine. Acronis still causes BSoDs though.

HD Tach also caused a BSoD simply by loading it, but after upgrading to a newer version that problem was completely fixed (?!?).

I took each hard drive out of the raid array and did a secure erase two different times now. I verified each drive with a chkdsk /r and did a few benchmarks on them to see if it would cause errors. As far as I can tell all the drives work fine and none of them were showing any weird issues.

Looking at the smart data I got a few Ultra DMA CRC Errors, which have climbed up to a total of 24 between all the drives. I swapped out the SATA cables to make sure they weren't causing it. I also reimaged the drives after finding this out so the image is fresh, all of the above still applies.

Besides these issues my system functions completely fine, so it's not completely fubard. I validated the image I installed from and it came up clean as well. Raid performance for the AMD 990X chipset sucks BTW. When it's in caching mode I essentially get SATA 2 speeds and when read ahead and the cache is disabled I get roughly 150MBps between all three drives. When adaptive readahead is on I get around 1100MBps, which isn't terrible. But that doesn't matter if the system turns into a giant corruption tube. I was only getting 800MBps on the Intel Z77 chipset I was using before.


This really has me stumped, there are only a handful of things I can think of. The computer I tested the SSDs on only has a SATA 1 port so the SATA 3 ports may be causing one or some of the drives to malfunction if they're on the edge of malfunctioning anyway. My motherboard may be bad or my southbridge is. This may be a bug pertaining to the 990X chipset. There are only a handful of boards that use that variation. Most boards use the 990FX chipset.

What I'm describing doesn't seem to be common place from all the googling I've done. If this is a AMD issue then there would be more posts, especially if it turns into a giant corruption machine. The motherboard I'm using is a newer model so this may only pertain to this particular model and brand. Compared to Intel chipsets and the Intel based Asus board I was using before, it doesn't look like Asus put much effort into this model. Reading Newegg reviews it appears a lot of the motherboard manufacturers don't put much effort into their AMD offerings now days. Looking at Gigabyte and ASrock boards, there are a lot of people complaining about trying to get the raid setup (but none about corruption once it's setup).

I'm leaning towards I have a bad board or this particular make and model wasn't properly debugged. It would seem at first that the SSDs are bad, but I really don't know after testing them individual. The computer works fine otherwise. Although I haven't used it long enough to see if this is a long term data 'rot' sorta thing. There really doesn't appear to be any problems when the board raid mode is set for caching instead of adaptive read ahead or disabled.

Update:

Disabling readahead and the cache completely allows me to successfully make a image through Acronis. Performance is dropped down to about 150MBps though. This just blows my mind.

Update 2:

Turns out the image was corrupt.
Last edited by Bensam123 on Wed Apr 10, 2013 11:14 pm, edited 1 time in total.
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Random BSoDs Raid 0 Asus M5A99X EVO R2.0

Tue Mar 26, 2013 10:35 pm

Just an update on this, as it turns out all three of my SSDs were in various stages of dying. Now one of my Samsung 830s isn't detected, another is flaky, and the third seems to be having similar issues it's just in earlier stages. I guess that's one bad mark in the 830s pristine profile.
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 12:06 am

I changed the topic name to more accurately describe my problem.

I received Samsung 830s back from RMA, which was pleasantly fast. The rep I talked to sent me a prepaid UPS label that was for overnight delivery. The package I received back was also aired to me. It took in total all of four days (not counting the weekend). I was pretty impressed.

However, I'm starting to learn things about the 830s that isn't too impressive. Apparently the 830s and 840s have issues with CRC errors, which is quite common, apparently so common that it's a part of normal every day use, you just don't normally notice them. Maybe occasionally you'll get a BSoD, but the actual system wide implications aren't that noticeable. The problem with this is I do Acronis image back ups of my 3xr0 array and this puts extended stress on the array, which eventually throws a CRC error and the entire array shuts down... Of course this results in a BSoD. I've found that if I turn 'ignore bad sectors' on, I no longer get BSoDs though. However, this does make me wonder how reliable that data is in that backup if it's throwing a CRC error and Acronis just ignores it. Doing a chkdsk /r also results in a BSoD at random points. Also running the normal 'benchmark' in HD Tune occasionally results in a BSoD.

I originally did not run into these issues on my Intel system with a very similar motherboard, the P8Z77-V LK. However, since I had three disks I ran all three disks on the SATA 2 ports instead of two on the SATA 3 and one on the SATA 2 in order to maintain uniformity. Thus the drives always ran at a slower speed and never up to their potential.

It seems as though these errors do not actually corrupt the data and one of the reasons I'm running into BSoDs is because they are in a raid array. This of course definitely increases my likeliness of corruption due to the array dropping out in the middle of a write (which has started to happen). Furthermore according to the smart data for my old drives I never got reallocated sectors or hardware ECC errors, only CRC Errors. This makes it seem as though Samsung has some issues with their SATA implementation.

I've been reading in general that 830 and 840 drives are very fincky. Some people stating that the cable makes a world of difference (I tried three different cables and this doesn't make any difference), some state that it depends on the motherboard, others that it depends on AMD or Intel. There may be a bit of legitimacy to that last part as Intel may be a lot better at handling data inconsistencies. I've tried four different boards, P8Z77-V LK, M5A99FX Pro R2.0, M5A99X Evo R2.0, and a old AMD system with SATA 1 ports on it. The old AMD system didn't run into any errors and neither did the Intel system. Since this was so sporadic I had thought that this may be caused by my motherboard and I returned it for a slightly different model, thinking it would make a difference, but it didn't.

I've done all the basic troubleshooting steps so I'm pretty darn sure it's not my system and even the refurbished drives I was sent exhibit these same issues. They apparently don't reset the smart data on the drives and one of the ones I received back had a whopping 140,782 CRC errors on it.

Looking around the web these CRC errors don't seem uncommon at all. There is one really big SSD reliability thread on xtremsystems, I haven't gotten a chance to look through it, but there are quite a few posts of 830s with CRC errors in them.

http://www.xtremesystems.org/forums/sho ... nm-Vs-34nm

On a related note, one of the topics they mention as some of the thread participants are actually writing the disks till they fail is none of the disks they've stressed have fallen into a 'read only state'.
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 12:12 am

FWIW, I've had 0 issues with my 830. Did you not reinstall Windows when switching from Intel to AMD? That's a no-no in my book.
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
BlackStar
Gerbil
Posts: 98
Joined: Fri May 11, 2007 3:38 am

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 3:28 am

I have encountered CRC errors when connecting the 830 to older Intel SATA controllers. As far as I can tell, this is a matter of hardware incompatibility - the same SSDs work perfectly fine on different controllers from both Intel and AMD.

Unfortunately, I haven't been able to find any kind of compatibility list, so it's a matter of luck and testing. The issue can be diagnosed by looking at the Windows system log for CRC errors (and it's always a good idea to monitor the log for a while after a new installation.)
 
anotherengineer
Gerbil Jedi
Posts: 1688
Joined: Fri Sep 25, 2009 1:53 pm
Location: Northern, ON Canada, Yes I know, Up in the sticks

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 7:47 am

I seen acronis and I cringed after hearing all the issues people have been having with the latest versions, however no personal experience myself.

I have a gigabyte 990x, with an AMD 955BE, and a Plextor M3S 256GB running happily for about 1 yr now with no issues, (sata 3.0 port and win 7 sata/ahci driver)

Usually the best thing with a new build is to completely format and start from scratch though.
Life doesn't change after marriage, it changes after children!
 
frumper15
Gerbil XP
Posts: 380
Joined: Mon Jan 18, 2010 3:25 pm

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 8:50 am

Do you have any other drives you can test out with your new setup? I just wonder if you're assigning the blame for these problems to the drives that had no issues for you on your intel setup but seem to have been nothing but problems on your AMD setup, even after replacing a number of them. Alternatively, have you tried running them not in RAID0, or maybe not in a 3-way configuration? Also, it sounds like you're restoring an image to this config instead of doing a clean install of windows. I'm assuming the image was initially created by the Intel setup that you have since replaced and simply imaged things over. I just wonder if you're focused on the drives being the source of the problem and haven't investigated changing other variables - after all, it sounds like you didn't have any problem with same drives on a different system in the same configuration.
i7-8086K | Z370 AORUS GAMING WIFI | 32GB DDR4-2400 | EVGA GTX 1080 Ti | 512GB 960 Pro | 27" Dell 2560x1440 Gsync | Fractal R6 | Seasonic Focus Plus 850W | Win10 Pro x64.
 
ryko
Gerbil Team Leader
Posts: 256
Joined: Tue Feb 27, 2007 3:58 pm
Location: new york

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 9:46 am

You need a fresh install. You can't go from one chipset to another and keep the same install. Sometimes if you are sticking to the same chipset manufacturer and just upgrading to a newer version you can slide by, but it is still not recommended. Going from intel to amd or vice versa will never work out without a fresh install. There are the intel matrix storage drivers, etc...that you need to get rid of and install the correct amd RAID drivers. Which probably explains your low speeds after the transfer.
 
absurdity
Gerbil Elite
Posts: 890
Joined: Sat Mar 02, 2002 7:00 pm
Location: VT

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 9:54 am

Does Samsung certify these drives for use in a RAID setup?
 
maxxcool
Gerbil Elite
Posts: 855
Joined: Thu Sep 12, 2002 8:40 am
Location: %^&*%$$
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 10:02 am

I do not see it mentioned but i'd test the drives individually out of the suspect system. unless you have been editing massive movies every day 24x7, or running a backup every other hour i do not see the flash cells 'dying' in under two years.

my immediate suspect is the amd raid controller. because to be honest although i love amd for being cheap and easy to use their raid chipset makes me want to hammer my sack with a wooden mallet until i taste the rainbow and smell sounds...

on various boards i have owned i have had to disable NCQ, or switch from "native acpi" to 'compatible' to get multiple sata drives to work, or switch to 'IDE' then raid-0 or raid 1 to get a array stable or detected.

so seriously, break the array and test each drive separately in either a intel box and or a amd box in native IDE mode, do a ''diskpart clean all''... if it fails to zero the drive then id call them dead... I am 99% sure you wont see the errors.

If you truly think there dead, mail them to me :D /evil!/ .. but im not so sure...
Cybert said: Capitlization and periods are hard for you, aren't they? I've given over $100 to techforums. I should have you banned for my money.
 
maxxcool
Gerbil Elite
Posts: 855
Joined: Thu Sep 12, 2002 8:40 am
Location: %^&*%$$
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 10:09 am

absurdity wrote:
Does Samsung certify these drives for use in a RAID setup?


probably not. but since it is not a mechanical drive it is less of a issue. however if he was running parity on one drive and not the two others it might explain how one drive may be near its wear level max and the other are not.

but i would stress my post above again. break the raid, test the drive outside the suspect box one drive at a time by doing a 'diskpart clean all'... if it cannot finish a zero write it really is dead...

testing in the raid setup can yield bad results in my personal experience with amd's oddball raid setup.
Cybert said: Capitlization and periods are hard for you, aren't they? I've given over $100 to techforums. I should have you banned for my money.
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 10:10 am

I REALLY don't think the RAID is the issue. You need to reinstall. It always causes problems. Seriously try that and let us know.
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
maxxcool
Gerbil Elite
Posts: 855
Joined: Thu Sep 12, 2002 8:40 am
Location: %^&*%$$
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 10:26 am

DancinJack wrote:
I REALLY don't think the RAID is the issue. You need to reinstall. It always causes problems. Seriously try that and let us know.


wait... did i miss some text? you did not do a clean re-install of the OS ?

Do that 1st , untill you remove the intel flotsam and garbage you can be seeing all sorts of bad api calls, bad dmi mapping, etc...

*Then* break the raid and test the drives individually outside the box...
Cybert said: Capitlization and periods are hard for you, aren't they? I've given over $100 to techforums. I should have you banned for my money.
 
Scrotos
Graphmaster Gerbil
Posts: 1109
Joined: Tue Oct 02, 2007 12:57 pm
Location: Denver, CO.

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 10:35 am

As an anecdotal data point, I have 8 840 Pro 256 GB drives set up in a RAID 1+0 with a HP SmartArray P400. I've put them in RAID 0, 5, and 6 and run iometer benchmarks on 'em in various configurations. I've transformed the array from one type to another and expanded it, changed the stripe size on the fly and had the array... mmm... reinitialize itself? So it was about 4 or 6 weeks of rebuilds and benchmarks.

This is typical of what the array diagnostics have for each physical drive:

Serial Number S12RNEACC72435F
Firmware Revision DXM04B0Q
Product Revision ATA Samsung SSD 840
Reference Time 0x00000567
Sectors Read 0x00000000223bc557
Read Errors Hard 0x00000000
Read Errors Retry Recovered 0x00000000
Read Errors ECC Corrected 0x0000000000000000
Sectors Written 0x000000000014907e
Write Errors Hard 0x00000000
Write Errors Retry Recovered 0x00000000
Seek Count 0x0000000000000000
Seek Errors 0x0000000000000000

There's more. I mean, a TON more of stuff, but the gist of it is that I haven't seen any errors that seem related to CRC. I got paranoid when you mentioned Samsung not being nice in a RAID so I went and checked.
 
flip-mode
Grand Admiral Gerbil
Posts: 10218
Joined: Thu May 08, 2003 12:42 pm

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 12:07 pm

It could be core parking! Did you disable that? :wink: 8)
 
Scrotos
Graphmaster Gerbil
Posts: 1109
Joined: Tue Oct 02, 2007 12:57 pm
Location: Denver, CO.

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 12:45 pm

Oddly enough...

viewtopic.php?f=6&t=87458
http://support.microsoft.com/kb/2775511

...includes this:

http://support.microsoft.com/kb/2646060

"An update that selectively disables the Core Parking feature in Windows 7 or in Windows Server 2008 R2 is available"
 
End User
Minister of Gerbil Affairs
Posts: 2977
Joined: Fri Apr 16, 2004 6:47 pm
Location: Upper Canada

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 1:05 pm

Scrotos wrote:
I have 8 840 Pro 256 GB drives set up in a RAID 1+0 with a HP SmartArray P400. I've put them in RAID 0, 5, and 6 and run iometer benchmarks on 'em in various configurations. I've transformed the array from one type to another and expanded it, changed the stripe size on the fly and had the array... mmm... reinitialize itself? So it was about 4 or 6 weeks of rebuilds and benchmarks.

Apologies for going off topic but I've been following your RAID 5 thread and this post pushed me into the realm of the super curious. What on earth on your doing with so much storage capacity?
 
Scrotos
Graphmaster Gerbil
Posts: 1109
Joined: Tue Oct 02, 2007 12:57 pm
Location: Denver, CO.

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 5:54 pm

It's only 1 TB usable space. Basically, we had a vendor come in and start whining about how they needed this and that and all kinds of junk for their custom software that handles stuff that a DOS program run locally on a generic Core 2 workstation could handle. Blah blah blah we'd need a db server and an app server to meet their "tier 4" requirement and because we had soooo many records (what, 5000? Seriously?) we'd need a ton of performance.

Boss went on vacation and right before then dumped it in my lap, "make it happen!" I ended up going with:

HP DL380 G5
2 x Xeon E5430 2.66 GHz (8 cores total)
32 GB RAM
P400 RAID controller with 512 MB of battery-backed cache
8 x 256 GB SSD in RAID 1+0

The cost of the server was around $500, the additional RAM was another $500 or so, and the SSDs were a bit less than $2000 if memory serves. OS was Win2K8 R2 Enterprise (2012 downgrade) for something like $750 and SQL Server 2012 Standard for around $750 as well. These clowns had write-ups for SQL Server 2005 and Win2K3 so the install process was an interesting bit of fumbling.

I also built a pretty identical server to replace our old SQL Server 2005, only difference being the drives are HP 146GB 6G DP 15K SAS drives instead of SSD. Performance is so much better than the server it'll replace, which wasn't really buckling under any strain anyway, that I hope it will last another 5 to 10 years.

The vendor had requirements for 150 GB OS drive, 170 GB database drive, 150 GB logging drive, and of course whatever the "application server" would need. When I originally specced out the system, I was assuming 146 GB SAS drives and RAID 6. So combining the app and db server would still be a piece of cake, space-wise. Then they started pissing me off so I said whatever, I'll put SSDs in there to shut them up. The older controller and backplane do limit the SSD performance and I will post benchmarks on that when I get a chance in the other thread, but it's waaaay faster than the mechanical drives.

I went with the 256 GB model mainly because the reviews I read pointed to that as a sweet spot for longevity and performance. Better than the 128 (and 64?) and about the same as the 512. Plus, I'd been exposed to them a bit as part of a slow deployment to our workstations.

So what do we NEED with all that space? Nothing, really. I will give them a cookie if they claim they are bottlenecked somewhere. Actually, looking at the 600+ separate tables they have in their db, I dunno, maybe they can find a way to make it suck. One product from one vendor that will cost in the tens of thousands. We run our entire business on a crummy old SQL server that handles way more records than what this vendor will and it's just fine. It boggles the mind.

The RAID 5 thread is what I ran into while researching proper storage solutions. Like any business, we have money to spend on stupid stuff but not on stuff that's the most appropriate. Probably we need a SAN and lots of virtualized servers or whatever, but what we get money for is old HP servers and whatever I can do to make them future proof as much as possible. Our automated fax server is running Win2K with a 4-port fax board that has 1 port broken. At some point we're going to have to get a new document archiving/imaging solution and I'd like to switch from our simple 2 TB mirror running on an old 3Ware RAID card to something proper, but I'm still feeling my way out on what that should be.

Sorry for the rant. Good luck to the OP!
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 7:49 pm

No, this isn't a fresh installation and no you don't need to do a fresh install when you change chipsets. This W7 installation has been through more upgrades then I can count, including at one time being Windows XP. I've never needed to reinstall as long as you clean out the old drivers and fix any problems along the way. The 'just reinstall' answer is generally always given by those that don't know how to fix a OS. That includes this installation being at one point on 2xr0 500GB Seagates, 3xr0 137GB Seagate SAS drives, and now 3xr0 128GB Samsung 830s. I've encountered no such errors along the way including when it was on the Intel chipset. It's completely irrelevant. And yes, I've tried a fresh installation just to see what would happen, which I still ended up with the same errors so I imaged it back. At one point I was running W8.

Drives were tested in two different systems outside of the raid array. The first three drives as I mentioned, one outright died, the second wouldn't detect on boot and sometimes just disappeared from the OS, and the third had CRC errors so I RMA'd it anyway as it was from the same batch. I've done all the troubleshooting protocols.

BlackStar wrote:
I have encountered CRC errors when connecting the 830 to older Intel SATA controllers. As far as I can tell, this is a matter of hardware incompatibility - the same SSDs work perfectly fine on different controllers from both Intel and AMD.

Unfortunately, I haven't been able to find any kind of compatibility list, so it's a matter of luck and testing. The issue can be diagnosed by looking at the Windows system log for CRC errors (and it's always a good idea to monitor the log for a while after a new installation.)


Yeah, from what I've been reading the drives in general seem quite finicky as to what you have them attached to. This may be in particular a issues with Asus and this chipset (although I already tried a different version of the 990 chipset). Check out the link I posted in my last post. This particular issues is easily reproducible by doing a chkdsk /r or running HD Tune 'benchmark' a couple times. The computer BSoDs and a CRC error is added to the SMART data.


Scrotos wrote:
Oddly enough...

viewtopic.php?f=6&t=87458
http://support.microsoft.com/kb/2775511

...includes this:

http://support.microsoft.com/kb/2646060

"An update that selectively disables the Core Parking feature in Windows 7 or in Windows Server 2008 R2 is available"


Yeah Scrotos don't even try, let the ignorant masses remain ignorant and then chortle to themselves when they think they've said something clever. That thread is still growing in page views.
 
maxxcool
Gerbil Elite
Posts: 855
Joined: Thu Sep 12, 2002 8:40 am
Location: %^&*%$$
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 11, 2013 11:37 pm

I still think the raid array when on the AMD chipset is the issue. But since you have validated the drives as dead and dying outside the suspect system further testing is moot until you have validated good new drives to install back for further testing.
Cybert said: Capitlization and periods are hard for you, aren't they? I've given over $100 to techforums. I should have you banned for my money.
 
Krogoth
Emperor Gerbilius I
Posts: 6049
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Samsung 830 Corruption Woes...

Fri Apr 12, 2013 3:16 am

Why are you running a RAID 0?

RAID 0 is just stupid. There's no excuse for it these days. A single SSD is fast enough for the vast majority of mainstream applications. The applications that need more I/O throughput typically fall into content creation and professional realm where you are using your system for real work. In that instance, youo can easily justify the expense of getting a decent SSD PCIe card that would easily outperform that RAID 0 array with none of the headaches.
Gigabyte X670 AORUS-ELITE AX, Raphael 7950X, 2x16GiB of G.Skill TRIDENT DDR5-5600, Sapphire RX 6900XT, Seasonic GX-850 and Fractal Define 7 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Samsung 830 Corruption Woes...

Tue Apr 16, 2013 9:48 pm

Yeah... this really has me stumped. It may just be trying to do raid on a AMD system (or the combination of Samsung 830s and the specific chipset I'm using), but I'd suspect more people would be posting issues about this (although AMD systems seem to be in the minority now days, let alone raid 0 arrays with SSDs).

I'm really quite leery about using the computer in it's current configuration, but it seems perfectly stable unless I try to do the above things I mentioned. I don't have the money to buy new SSDs (nor do I want to go back to mechanicals).


Cause I can Krogoth and for the same price as buying a 256GB drive I could buy three 128s at the time. I really don't care about optimizing for a specific workload, I'm a enthusiast and I do what I want because it makes me feel good on the inside and makes my epeen grow three inches. If I built this computer for my g'ma or brother or something, most definitely I would probably have a small SSD in there (probably using intel RST). Maybe I wouldn't even have a SSD, but they aren't me.

I've seen a nothing but issues with PCIE based SSDs (reading Newegg reviews as well as browsing the net). I wouldn't spend money on one. They cost no where close to the $270 I paid for my three Samsung 820 128s.
 
BIF
Minister of Gerbil Affairs
Posts: 2458
Joined: Tue May 25, 2004 7:41 pm

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 12:50 am

Krogoth wrote:
Why are you running a RAID 0?


Bensam123 wrote:
Cause I can Krogoth and...


But can you? Really? With all these problems, maybe everything that comes after the word "and" in your sentence is moot. "Because I can" is a good reason to do a lot of things in life. Climb a mountain, bungee jump off a bridge, go skydiving.

Maybe in this case, the minor-to-zero improvement is just not worth the trouble and the drives that keep coming up bad. Stop the madness. Simplify! :P
 
FubbHead
Grand Gerbil Poohbah
Posts: 3482
Joined: Mon Apr 08, 2002 9:04 pm
Location: Borås, Sweden

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 12:55 am

Tried going Windows software-raid instead?
Amiga 1200, 68020@28MHz, 4MB+2MB RAM, Conner 80MB harddrive
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 6:46 am

At one time and the overhead for it is ridiculous... I wouldn't use windows raid for anything... >>

And I don't know BIF... perhaps, perhaps not... I don't think I'm trying to move a mountain by putting three drives into a raid 0 array. XD
 
DancinJack
Maximum Gerbil
Posts: 4494
Joined: Sat Nov 25, 2006 3:21 pm
Location: Kansas

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 7:59 am

Sorry if you did this and I missed it, but have you tried a reinstall? I know we were harping about this and you didn't think it would matter, but have you at least tried it?
i7 6700K - Z170 - 16GiB DDR4 - GTX 1080 - 512GB SSD - 256GB SSD - 500GB SSD - 3TB HDD- 27" IPS G-sync - Win10 Pro x64 - Ubuntu/Mint x64 :: 2015 13" rMBP Sierra :: Canon EOS 80D/Sony RX100
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 5:39 pm

Yeah, I mentioned I tried a fresh install of W8 at one time (and one of W7), it still does it so I imaged it back.

A bit of a update to this. So I accidentally left disk defragmentor on last night (which was still set from sending in my SSDs for RMA, I had a mechanical in) and it of course BSoD'd. Upon reboot I decided to run chkdsk to see if it caused some corruption and I had to run a /f. Upon doing so I had to have had maybe a thousand or so errors on the drive. I decided to reimage it right away from a image I had made the night before.

On my 3570k I did a defragmentation one day to see what would happen (I know it uses up write cycles and changes largely nothing) and it didn't cause a BSoD at that time or any errors. Once again the drives were only operating at SATA 2 speeds though.

Now, something else that's weird is I have the array setup to do a patrol read, which should catch bad sectors. However I've done a patrol read on about five different occasions over the last few months and it's never picked up a bad sector. It doesn't cause the computer to BSoD either even though it's scanning all the sectors. Running a chkdsk /r causes the computer to BSoD once again. This I think relates to the amount of disk activity. A patrol read is a very minimal background task (it takes about two hours and maybe utilizes 5% of the drive).
 
Krogoth
Emperor Gerbilius I
Posts: 6049
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 7:42 pm

Just stick with JBODs or single disks if you want to get high capacity with SSDs on the cheap.

RAID 0 is not worth it at all. Too many headaches to warrant the benefits.
Gigabyte X670 AORUS-ELITE AX, Raphael 7950X, 2x16GiB of G.Skill TRIDENT DDR5-5600, Sapphire RX 6900XT, Seasonic GX-850 and Fractal Define 7 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B
 
Bensam123
Gerbil Elite
Topic Author
Posts: 990
Joined: Wed May 29, 2002 12:19 pm
Contact:

Re: Samsung 830 Corruption Woes...

Wed Apr 17, 2013 11:37 pm

Thanks for the insight Krogoth, I've been using raid for over a decade... I know the ups and downs of using it. Still trying to figure out what's causing this problem though. I'm possibly thinking a PSU issue? I haven't really had any PSU issues though... That may make sense as I'm using a 8350 now and it draws more juice then a 3570k... adding the drives on the SATA 3 they may draw more juice as well. That's kinda reaching for a answer though.
 
Krogoth
Emperor Gerbilius I
Posts: 6049
Joined: Tue Apr 15, 2003 3:20 pm
Location: somewhere on Core Prime
Contact:

Re: Samsung 830 Corruption Woes...

Thu Apr 18, 2013 12:00 am

I don't think PSU is the issue here. I'm willing to bet the problem resides with firmware on the controller and/or SSDs. RAID controllers were meant to work with HDDs not SSDs, likewise customer-grade SSDs aren't meant to put inside a RAID environment.

The first-generation of customer-grade SSDs had interesting sets of issues dealing with sleep mode and abrupt power loss. Some of them resulted in irrevocable corruption of the file system. Most of the problems were addressed with firmware updates and occur less frequently on newer generation of SSDs.

I think we are seeing something similar here with your setup. Again, RAID 0 is just a bad idea outside of a scratch disk. If you need RAID some of kind. I wouldn't settle any less than RAID 1 or 10.
Gigabyte X670 AORUS-ELITE AX, Raphael 7950X, 2x16GiB of G.Skill TRIDENT DDR5-5600, Sapphire RX 6900XT, Seasonic GX-850 and Fractal Define 7 (W)
Ivy Bridge 3570K, 2x4GiB of G.Skill RIPSAW DDR3-1600, Gigabyte Z77X-UD3H, Corsair CX-750M V2, and PC-7B

Who is online

Users browsing this forum: No registered users and 1 guest
GZIP: On