cr*p! sata controllers dying, need replacements!

[tmp31416]

well, doing some non-renovation stuff yesterday evening, i was moving files around, tidying up some shares and whatnot, uploading some recently acquired videos (mostly tv shows that aren't broadcasted anymore in my area)... only to discover a locked up server when i returned to it, after starting a massive upload ~half an hour before that.  as if washing the dishes was bad karma or something.

 

it turns out one of my """trusty""" syba 3132-based sata controllers gave up the ghost.  syba / "io test" sy-pex40015.  its replacement (i bought two such cards to replace another one that went bad previously) is doa, not even showing up in the windows' device manager (this is the replacement to the replacement -- what luck!).

 

three defective cards out of total of 5: clearly a sign from ${deity}, i better find something else asap.

 

googling during my lunch hour today, i found these options:

* rosewill rc-211

* highpoint rocketraid 620 (there seems to be two versions of this card, one more expensive than the other: hello?)

* highpoint rocketraid 2300

 

unless i want to replace my motherboard or get rid of the video card (pci express ati 4350, passive), i have to use a pci-e 1x slot.

either i simply swap the defective syba for a rosewill or i replace all my syba cards with a pair of '2300, but this would be expensive (newegg.ca, which too often ships  defective parts to non-usa customers then charging "restocking fees").

 

i am sure i am missing one+ option(s), so, feedback, anyone?

 

btw, silly question: is there such a thing as a pcie 1x video card for simple console duties?  never saw such a beast, but what do i have to loose by asking?

 

thank you in advance for any feedback provided.

 

cheers.

 

 

[tmp31416]

replying to myself as an update:

 

i have found an available pcie-1x video card, to my great surprise.

it's a powercolor 4350-based card, passive; model # ae4350 512md2-h for those interested.

anyone seen / used this?  good, bad?

 

a local store appears to be selling it, will have to ring them tomorrow to confirm.

 

this means that i now have an available pcie x16 slot (the one for video, on my ga-ma770-ud3 motherboard) if this video card can be made to work with unraid.

the question now shifts to: anyone has a suggestion for the most cost effective unraid-compatible sata card that can support 4-6 drives?

 

cheers.

[WeeboTech]

MATROX MILLENNIUM G550 PCI-E X1

I've purchased a few of these used off ebay for $20-$40.

[WeeboTech]

the question now shifts to: anyone has a suggestion for the most cost effective unraid-compatible sata card that can support 4-6 drives?

 

I've used the Rosewill RC-218 with success (4 drives).

http://www.newegg.com/Product/Product.aspx?Item=N82E16816132018

 

many others have used the Supermicro AOC-SASLP-MV8 with success.

Although it will take some time finding a vendor and you need specific cables.

It does have 8 ports and is cost effective.

http://www.supermicro.com/products/accessories/addon/AOC-SASLP-MV8.cfm

 

I think by the time you buy the rosewill, that money could have been invested into the AOC-SASLP-MV8.

[calvinandh0bbes]

many others have used the Supermicro AOC-SASLP-MV8 with success.

Although it will take some time finding a vendor and you need specific cables.

3 in stock as of this post....

 

http://www.provantage.com/supermicro-aoc-saslp-mv8~7SUP918N.htm

[xamindar]

3 dead cards? In years of using computers I have only had one thing fail besides hard drives. A motherboard with defective capacitors that all popped. Are you sure something is not causing this? Bad power on the pci bus? Too much heat in the case?

[WeeboTech]

3 dead cards? In years of using computers I have only had one thing fail besides hard drives. A motherboard with defective capacitors that all popped. Are you sure something is not causing this? Bad power on the pci bus? Too much heat in the case?

 

Once I had a bad power supply fry 90% of my computer.

The only thing that survived was the ram and intel CPU.

 

At the time I did not know it was the PSU and I thought the motherboard died.

 

When I replaced the motherboard and turned the machine on, I said to my friend,  in a joking manner.

OK, now backup, because we don't want the sparks to get you... She backed up, but only a lil.

When I turned on the PSU, every diode(not led, regular diode) on the motherboard lit up, turned orange and burned out with a sizzle and crack.

She backed up and screamed and said, "I'm never going to build computers, no way."

All I could do was laugh, then sigh when I realized what happened.

 

A bad PSU can fry components really fast.

Not pulling the plug when you install a card in a slot can also zap a card.

I've known a few friends who did this with memory and add in cards.

[Joe L.]

When I replaced the motherboard and turned the machine on, I said to my friend,  in a joking manner.

OK, now backup, because we don't want the sparks to get you... She backed up, but only a lil.

When I turned on the PSU, every diode(not led, regular diode) on the motherboard lit up, turned orange and burned out with a sizzle and crack.

She backed up and screamed and said, "I'm never going to build computers, no way."

All I could do was laugh, then sigh when I realized what happened.

So that's how you got the nickname of "Sparky"  ;D

[dogmanky]

dangling wires inside a case that is running can also create an exciting moment if they touch the wrong thing.

 

[tmp31416]

update:

 

the supermicro AOC-SASLP-MV8 is indeed the least expensive 8 port card out there, but as WeeboTech said, it is nearly impossible to find.

calvinandh0bbes pointed to provantage as a source, but delivery cost + the inevitable extortion fee asked by dhl (same as ups, etc.) made buying it there a non-option.  (companies like ups, fed-ex, dhl and so forth can and will too often ask for all sorts of unreasonable fees for cross-border shipments between the usa and canada)

i have not found a store in canada that stocks this controller, so i will have to hunt for used items.

unless anyone in "the great white north" knows one that has them in stock, of course.

 

the "totally doa" card did not even make it in my server.  i simply put it in another box (running xp) that i use to flash the firmware from raid to non-raid version, and it simply did not show up in the device manager.  i tried re-seating the card to no avail.

 

surprise: on a whim, i put the card that locked up my server in the other machine (mentioned above) and it did behave correctly (in appearances).  so tonight, i put it back in the server and lo!  it's working for now.  the server is currently busy doing a parity check and i am right now praying to anything that will listen for the server to finish the check asap so that i can shut it down before it locks up again.  it is only at 4.5% and some of the drives are hitting 34C.

 

i am unsure at what exactly happened, though it is certain that i do have a doa syba card.  why a third one is going "marginal" is still unclear.  the thought of a sub-par psu did cross my mind, as well as a defective motherboard and so on, but:

- a 600w ocz modxtream is not exactly some no-name rubbish

- this motherboard has been running for roughly 6 months without a hiccup (except when the first syba card to go bad refused to let it complete its post): if the motherboard was defective, i think it is reasonable to expect the "shake-down period" to be within 6 months, no?

- if you look at the lounge's "pimp my rig" thread, you will see i did not exactly skimp on cooling.  i might have swapped the 5-in-3 drive cage's stock fans for units with less cfm, but i do not think that my drives are burning up (at 7.6% right now and the drives are now between 30C and 35C (oops, i've never seen them this warm  - maybe i'll put back the stock ones when i move my server in the basement???)).

- though there is a possibility that the pcie-1x slot "behind" the 16x slot might be going bad, but i am not sure how or why.

 

as it stands, i am hoping that swapping the 3 syba cards for a single AOC-SASLP-MV8 might lessen the thermal/electrical load, even by just a little bit.  also, a ~100usd card might be of better quality than 17usd ones, which might make the biggest difference.

 

of course, any other hypothesis on what happened and suggestions for improving things are welcome.

 

cheers.

 

 

[WeeboTech]

dangling wires inside a case that is running can also create an exciting moment if they touch the wrong thing.

I once plugged the floppy connector in wrong. In a few moments I smelled burning plastic.. saw smoke, then saw the wires turn read with the plastic dripping off. I was surprised the PSU did not turn off.

[WeeboTech]

update:

 

the supermicro AOC-SASLP-MV8 is indeed the least expensive 8 port card out there, but as WeeboTech said, it is nearly impossible to find.

calvinandh0bbes pointed to provantage as a source, but delivery cost + the inevitable extortion fee asked by dhl (same as ups, etc.) made buying it there a non-option.  (companies like ups, fed-ex, dhl and so forth can and will too often ask for all sorts of unreasonable fees for cross-border shipments between the usa and canada)

i have not found a store in canada that stocks this controller, so i will have to hunt for used items.

unless anyone in "the great white north" knows one that has them in stock, of course.

 

the "totally doa" card did not even make it in my server.  i simply put it in another box (running xp) that i use to flash the firmware from raid to non-raid version, and it simply did not show up in the device manager.  i tried re-seating the card to no avail.

 

surprise: on a whim, i put the card that locked up my server in the other machine (mentioned above) and it did behave correctly (in appearances).  so tonight, i put it back in the server and lo!  it's working for now.  the server is currently busy doing a parity check and i am right now praying to anything that will listen for the server to finish the check asap so that i can shut it down before it locks up again.  it is only at 4.5% and some of the drives are hitting 34C.

 

i am unsure at what exactly happened, though it is certain that i do have a doa syba card.  why a third one is going "marginal" is still unclear.  the thought of a sub-par psu did cross my mind, as well as a defective motherboard and so on, but:

- a 600w ocz modxtream is not exactly some no-name rubbish

- this motherboard has been running for roughly 6 months without a hiccup (except when the first syba card to go bad refused to let it complete its post): if the motherboard was defective, i think it is reasonable to expect the "shake-down period" to be within 6 months, no?

- if you look at the lounge's "pimp my rig" thread, you will see i did not exactly skimp on cooling.  i might have swapped the 5-in-3 drive cage's stock fans for units with less cfm, but i do not think that my drives are burning up (at 7.6% right now and the drives are now between 30C and 35C (oops, i've never seen them this warm  - maybe i'll put back the stock ones when i move my server in the basement???)).

- though there is a possibility that the pcie-1x slot "behind" the 16x slot might be going bad, but i am not sure how or why.

 

as it stands, i am hoping that swapping the 3 syba cards for a single AOC-SASLP-MV8 might lessen the thermal/electrical load, even by just a little bit.  also, a ~100usd card might be of better quality than 17usd ones, which might make the biggest difference.

 

of course, any other hypothesis on what happened and suggestions for improving things are welcome.

 

cheers.

 

I would not worry about drive temps if they stay under 40c.  If they reach mid 40s (45c) I would start to shutdown.

In the summer my drives are always around 40-44c. It's just that hot outside so there is no way they could be cooler.

 

I'm starting to question the syba cards themselves. Another member had strange issues with them.

 

A lockup is usually hardware to software related in regards to "speed" or "communications" or something is starved.

Power problems usually manifest as intermittant failures, intermittant reboots or hard locks when the machine is really taxed.

 

I would look to simplify the hardware layout and go with the name brand supermicro card.

 

How many drives do you have?

 

[tmp31416]

I would not worry about drive temps if they stay under 40c.  If they reach mid 40s (45c) I would start to shutdown.

In the summer my drives are always around 40-44c. It's just that hot outside so there is no way they could be cooler.

 

I'm starting to question the syba cards themselves. Another member had strange issues with them.

 

A lockup is usually hardware to software related in regards to "speed" or "communications" or something is starved.

Power problems usually manifest as intermittant failures, intermittant reboots or hard locks when the machine is really taxed.

 

I would look to simplify the hardware layout and go with the name brand supermicro card.

 

How many drives do you have?

 

re. the syba card:

 

this is the model i ended up getting:

http://www.newegg.com/Product/Product.aspx?Item=N82E16816124030

http://www.monoprice.com/products/product.asp?c_id=104&cp_id=10407&cs_id=1040702&p_id=2530&seq=1&format=2

instead of the better rated one that i wanted initially:

http://www.newegg.com/Product/Product.aspx?Item=N82E16816124017

i thought shipping the cards back was too much trouble, thinking "might as well keep this, it's based on the same chipset, specs are pretty much the same, how different can they be?"... 

 

i have 8 drives at the moment:

3 WD10EADS

2 WD5000AAKS

3 WD10EARS

6 drives are connected to the motherboard's sata connectors, 2 drives are connected through the syba cards (the last card is connected to a pair of empty drive bays (Kingwin KF-1000-BK), see the pics of my rig in the "pimp my rig" thread).

 

...and the server just hanged again (past the 35% mark in the parity check).  it could not be overheating, as the drive temps were dropping last i checked (only one was at 35C, from two or three), about 15 minutes ago.

 

it is not the first time i have a parity check with all 8 drives running, and the previous time, it completed without a hitch.

 

if this means anything to anyone, this time, like two days ago, all fans were running and the video card was, huh, "broadcasting" a blank screen when i came back.  but all drive activity was stopped (to state the obvious, before i stepped away from the box, all 8 drive activity lights were on).

 

and to reiterate, the suspicious card has no drives connected to it, so no storage activity could have been going through it.

 

i am a bit uneasy to turn the machine back on, as it is surely going to die on me again before completing a parity check.  i mean, to how many such failures can my data survive?

 

talk to you tomorrow...

[gfjardim]

DocBlock and I bought the Supermicro AOC-SASLP-MV8 from this guy:

 

http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=300360847366&ssPageName=STRK:MEWNX:IT

 

 

[WeeboTech]

Hmmm.. what would I do at this point.

 

I might take the 2 500 GB drives offline from the array. Unplug the power and see how the system works with 6 drives.

I do not think it's heat. there are plenty of fans.

I do not think the power supply is underpowered. it has 2 rails of 25A each.

Turn off everything in the bios that you can.  ide,serial(if not needed), firewire.

I've had lockups like this with heavy I/O and had to change boot options in the syslinux.cfg.

Stuff like noacpi and other parameters.

I would consider

1. disconnect and reconnect each sata cable on each end. Insure tight connectivity.

2. pull out and pop in each drive in the removable units.

3. checking the smartlogs on every drive before the parity check.

4. checking the smartlogs on every drive after the parity check.

 

 

>> (the last card is connected to a pair of empty drive bays (Kingwin KF-1000-BK),

Disconnect the cables from the controller or take the controller out temporarily.

I've seen issues like this with dangling connections and picking up stray signals and sudden lockups.

I'm sure there will be comments about it, but I've seen it before.

[tmp31416]

 

>> (the last card is connected to a pair of empty drive bays (Kingwin KF-1000-BK),

Disconnect the cables from the controller or take the controller out temporarily.

I've seen issues like this with dangling connections and picking up stray signals and sudden lockups.

I'm sure there will be comments about it, but I've seen it before.

 

whilst i'm not too sure about removing drives in the array's current state, i was already thinking of turning things off in the bios (not much left to, though), pulling out the controller that is causing me grief, telling unraid where the drives are (this card is the *last one scanned* by the bios during post and yet it will cause drives to move around, which, in my mind, does not make sense) and let it complete a parity check.  yes, the check kicks in automagically the minute all drives leave the "unformatted" state.  i am 90% sure it could complete the check with these ~minor changes.

 

once the parity checks completes, i will shut down the box until i get all the replacement parts (new graphics controller and that supermicro sata controller).  i'm not sure what else could be replaced / should be replaced...

 

btw, thank you for confirming the psu is not junk.  that's one thing less on my mind.

 

 

[WeeboTech]

You can try some of the kernel boot options as in this post.

 

http://lime-technology.com/forum/index.php?topic=6399.msg63164#msg63164

http://lime-technology.com/wiki/index.php?title=Boot_Codes

[xamindar]

(this card is the *last one scanned* by the bios during post and yet it will cause drives to move around, which, in my mind, does not make sense)

Last drive scanned by bios does not mean it is the last card loaded by Linux as well. Unraid probably loads the modules for your cards in a different order than the bios scans them. Nothing wrong with that and quite normal.

[tmp31416]

quick update: doing a quick check on my server, after i removed the problematic sata controller out of it and turned it on, and it is still going strong doing a parity check.

now at 51.5% and all appears to be fine.

 

did not change anything else, as everything that could be turned off in the bios was, already.

 

this singles out the SY-PEX40015 (http://www.newegg.ca/Product/Product.aspx?Item=N82E16816124030) as an apparently undesirable component. fyi.

 

to be continued...

 

** update #2:

 

the parity check completed successfully, the server is still alive and responding. will now check the filesystem to assess the damages (missing files, etc).

 

as indicated elsewhere, i have found a used AOC-SASLP-MV8 and will replace the SY-PEX40015 cards with it, as i am now certain they are not desirable to have in my machine.  do i have to upgrade from unraid 4.5 to make use of it?

 

cheers.

 

p.s.: the supermicro card appears to be such a good deal that i cannot understand why its manufacturer is not flooding the market with them.  unless it is a "loss leader" to get people buying other supermicro gear?!?!

[spinbot]

i have not found a store in canada that stocks this controller, so i will have to hunt for used items.

unless anyone in "the great white north" knows one that has them in stock, of course.

 

I am in the same boat as you.  I'd like to make the purchase as well as a couple new 2TB drives before the HST comes into effect in B.C./Ontario so I can save myself the 8% as I usually do when buying from out West.

 

I am going to try a different source for the controller tomorrow and if I have any luck, I will report back so other Canadian's might be able to get one.

 

I hate when I decide on something to get, then can't get it!

[tmp31416]

hello, world --

 

a very delayed update:

 

* i eventually found a not very used supermicro card, bought and installed it in my server.  i do wish these cards would be more readily available, esp. north of the usa, as they really are a very good deal for the functionality offered.

* of course, i removed all the syba sata controllers from the box.  better be safe than sorry.

* after installing and booting 4.5.0, nasty messages of doom and gloom on the console (and the webgui) motivated me to upgrade to 4.5.4; once that was done, i only had to re-assign the drives and voila! my array was up and running!

* dealing with the thinner sata cables that came with the supermicro AOC-SASLP-MV8 made me wish i could find more of these, but of the "1-to-1" type instead of the "1-to-4" type.  such cables do help make the case's interior tidier, even if it's a quick-n-dirty job.  anyone knowing a source for "skinnier" sata cables would make quite a few around here happy if {s}he would post the coordinates of the supplier.

 

so all is fine with my server, afaik.

 

cheers.

 

p.s.: ...and anyone with an unused slot cover with a ps/2 connector on it (attaches to leads on the motherboard) can wave at me, i'm still looking for one of those.  willing to pay for it (within reason, of course).

 

p.p.s.: something like this -- http://mycableshop.com/itemdesc.asp?ic=PLATE6F&eq=&Tp=

 

edit, 2010.07.07:  found a source for those elusive backplates i was looking for.  startech.com, look for "plate6f".  more expensive than it should have been, but at least i will get it and have my server finally "luggable" (the reset switch is currently outside the case, pretty much vulnerable).

 

 

[poofyhairguy]

Please let us know if you have any speed differences between the new card and the old Sil 3132 cards. I use lots of the Rosewill Sil3132 cards in my servers, and if its faster I will upgrade.

 

Thanks!

[tmp31416]

Please let us know if you have any speed differences between the new card and the old Sil 3132 cards. I use lots of the Rosewill Sil3132 cards in my servers, and if its faster I will upgrade.

 

Thanks!

 

hello --

 

the biggest difference i see between the "old" config and the "new" one is that my server is not locking up, to my great relief.  besides that, i did notice that some mass copy command could take more time to start up (not sure why), but once it did, same perceived speed as with the cheaper syba cards.

 

i do not think the supermicro card would bring a dramatic increase in performance to a server -- i could be wrong, though.  it just makes it more practical to add more drives to said server (i.e., 1 card instead of 3 or 4...), i'd say.  

 

cheers.

 

[comfox]

If you are still looking for the Supercom AOC-SASLP-MV8 and for a place in Canada that sells it you can special order it from www.bewawa.com.  They are based in Toronto, great company to deal with.  They say it takes 5-7 days to order it in.

[tmp31416]

If you are still looking for the Supercom AOC-SASLP-MV8 and for a place in Canada that sells it you can special order it from www.bewawa.com.  They are based in Toronto, great company to deal with.  They say it takes 5-7 days to order it in.

 

well, since my current motherboard has only one pcie slot > 1x (gigabyte ga-ma770-ud3), i cannot use another one for the foreseeable "short" term (famous last words...).

 

for those outside the gta (if you are in ottawa, or montréal, etc.), oemexpress.com has allied itself with bewawa, so it might become easier to obtain such a card north of the border.

 

cheers.