Are all the 8-port cards the same?

Is there any reason to go with one 8-port controller card over another or are they all basically the same? I'm on 5.0 and would like to start sleeping the server and waking via magic packet. Parity is 4TB but all other drives are <=2TB. My board has 2xPCI-e x8 slots. I also have Amazon gift cards so buying from AZ is preferred. I have been unsuccessful at preventing random parity sync errors that always occur at different addresses and different amounts each parity check.

m1015

m1115

AOC-SASLP-MV8

AOC-SAS2LP-MV8

9211-8i

I suspect the performance will be essentially the same on all of these except the SASLP-MV8, which only has an x4 interface.    All of the other cards are x8 cards, so they'll have the same bandwidth, and should all be able to sustain full speed transfers with 8 drives with no problem.

I suspect the performance will be essentially the same on all of these except the SASLP-MV8, which only has an x4 interface.    All of the other cards are x8 cards, so they'll have the same bandwidth, and should all be able to sustain full speed transfers with 8 drives with no problem.

Thanks. Do you know if they all support sleep? I know in the past some cards wouldn't come back from sleep or would hang the system. I'm running a non-correcting check and am at 8 sync errors and it's only at 7% so far. I just want the errors to go away, I'm sick of seeing them. I can't even remember the last time I ran a correcting parity check due to the random errors. It would be great if I could borrow one of these but I don't see that happening.

The only way to get rid of the sync errors is to run a correcting check 

... as for sleep => I agree a lot of add-in cards don't properly support S3, but unless the specs happen to explicitly note support for it, I don't know any way to confirm it other than to just try it.    Perhaps some who use the individual cards can comment here if they're using S3.

The only way to get rid of the sync errors is to run a correcting check 

Not if the sync errors never match between checks Nor are they ever the same amount of errors. Something other than the data on the drives is causing the errors.

... as for sleep => I agree a lot of add-in cards don't properly support S3, but unless the specs happen to explicitly note support for it, I don't know any way to confirm it other than to just try it.    Perhaps some who use the individual cards can comment here if they're using S3.

Yup, I'm hoping others chime in with their experience.

... the sync errors never match between checks ...

Wow.  That's a very interesting symptom.  Have you THOROUGHLY tested your memory recently?  (i.e. 12-24 hours of Memtest)

Could be power supply.  Could be cross-talk between SATA cables, especially if they are bundled together for neatness.  Not all SATA cables are created equal, and some (probably cheaper ones) would be more prone to this than others.  Just a couple of other avenues to explore. 

But I would also go for the 24 hour Memtest as a very first step.  Any errors reported at all need to be addressed before doing anything else.

Looking at the component list, I'd definitely suspect the power supply.  While Corsair makes some very nice PSU's, the CX series are not among them -- this is their low-end series, and I've seen several issues with them in various systems.    Certainly wouldn't hurt to try a higher quality unit.

Looking at the component list, I'd definitely suspect the power supply.  While Corsair makes some very nice PSU's, the CX series are not among them -- this is their low-end series, and I've seen several issues with them in various systems.    Certainly wouldn't hurt to try a higher quality unit.

Inconsistent parity checks certainly point to memory issues. 24 hr memtest is the first step. Cable crosstalk I would say is next, although in my experience rare. The PSU seems unlikely to be the culprit on this type of issue, although I agree with Gary it is not a component I would recommend. A possibility, although very rare, is bad cache memory on one of the disks. If both memory and cable adjustments don't resolve your issue post back for other suggestions.

Could be power supply.  Could be cross-talk between SATA cables, especially if they are bundled together for neatness.  Not all SATA cables are created equal, and some (probably cheaper ones) would be more prone to this than others.  Just a couple of other avenues to explore. 

 

But I would also go for the 24 hour Memtest as a very first step.  Any errors reported at all need to be addressed before doing anything else.

When I first began having this problem I ran Memtest for 72+ hours without any errors. I was really hoping it would be as easy as bad memory. I recently replaced all the cables with shorter ones and unbundled them. I ordered over twice the cables I needed so I went through a few of them and the sync errors still persist. I even removed all drives from their hot-swap cages to no avail. One thing of note is that I tried to run prime95 and immediately it crashes citing that it ran out of memory. And no matter what I set the memory usage to it always immediately crashes due to memory running out. Though I always have almost all 2GB memory free at all times. This is a pretty vanilla server, there are no add-ons running and no cache drive, just data disks and a parity.

Is the PSU thread updated? I originally ordered the CX PSU because it's in the recommended list, that's not the case anymore? What do you guys recommend? I will never run more than 10 drives, especially with 4TB+ drives readily available. I'm currently running 7 green drives. I have gift cards to Amazon so I'll have to buy it there.

Also, before I added the Sil3132 a few days ago I was getting around 5 sync errors per check, now I'm getting over 50.

... as for sleep => I agree a lot of add-in cards don't properly support S3, but unless the specs happen to explicitly note support for it, I don't know any way to confirm it other than to just try it.    Perhaps some who use the individual cards can comment here if they're using S3.

 

Yup, I'm hoping others chime in with their experience.

...I can report that the M1015 w/ & w/o expander is running fine with S3.

As a side  note, I'd like to add that S3 support is not a common feature in all server boards.

 

...I can report that the M1015 w/ & w/o expander is running fine with S3.

As a side  note, I'd like to add that S3 support is not a common feature in all server boards.

I'm not using a server board

What board are you using?

The Corsair TX series or above (TX, HX, AX), or a Seasonic RM or above (M12, X) are all fine.  You don't need more wattage ... and you MAY not even need a better supply; but I've definitely seen cases (3 that I remember) where someone had done a build with a CX unit, and as part of the troubleshooting I swapped in a spare Corsair HX supply I have on my bench -- and all the issues disappeared.    They then just bought better power supplies and all was well.

For a system the size you've indicated, this would be a good choice:

http://www.amazon.com/Seasonic-SSR-450RM-450-Watt-certified-Supply/dp/B00918MQ8G/ref=sr_1_1?ie=UTF8&qid=1397842770&sr=8-1&keywords=450RM

I have an old Ultra PSU, would it even be worth swapping that in to see if errors disappear?

Also, since the last added drive (or Sil3132 card) seems to be creating more errors is there a way to remove that drive, do a parity check, and then reinsert that drive without losing data or rebuilding the aray while the 7th drive is out?

 

...I can report that the M1015 w/ & w/o expander is running fine with S3.

As a side  note, I'd like to add that S3 support is not a common feature in all server boards.

 

I'm not using a server board

 

What board are you using?

I used to run a Supermicro X8SIL-F which did *not* support S3..

Since then I have retired my unRAID licenses because I needed a Full-Disk-Encryption Layer in my NAS.

I am now running Linux Debian Wheezy with ZFS on an ASUS AM3+ desktop board (this runs with ECC and an Opteron 3350HE with AES)..getting 112+MB/sec constant read and write with full encryption...couldn't be happier.

I re-used my M1015s and my Intel RES2Cv240 expander..S3 and resume is working perfect.

I have an old Ultra PSU, would it even be worth swapping that in to see if errors disappear?

Wouldn't hurt to just swap the PSU and see if anything changes.  Although if you're swapping a low-quality unit for another low-quality unit then if it doesn't help you still couldn't be sure that's not the issue.    I'd swap it just to see.

... is there a way to remove that drive, do a parity check, and then reinsert that drive without losing data or rebuilding the aray while the 7th drive is out?

No ... once you remove a drive the system is "at risk" (not protected from any other failures).    What you COULD do is remove that drive; do a New Config; and let a new parity sync run;  then do a parity check after that to see if it works okay.    All of your data would remain intact -- it just wouldn't be parity protected until the sync was done -- but it sounds like it's very iffy whether it's truly protected now anyway.    You'd have to do another New Config to add the other drive back in with its data intact (if you just added it, UnRAID would clear it).

Question:  When you're getting these random sync errors, do any of the drives show errors?

Question:  When you're getting these random sync errors, do any of the drives show errors?

Nothing in the syslog. I have had ATA reset errors, but that was due to a flaky cage and has since been fixed.

I've talking about errors shown in the Errors column on the main Web GUI.

I've talking about errors shown in the Errors column on the main Web GUI.

I've never seen errors there.

Definitely interesting.  You've thoroughly tested the memory; and swapped SATA cables enough that it's not likely that's the issue ... so it's almost got to be either a bad SATA port, or a power issue.    I'd try the power supply next.

Definitely interesting.  You've thoroughly tested the memory; and swapped SATA cables enough that it's not likely that's the issue ... so it's almost got to be either a bad SATA port, or a power issue.    I'd try the power supply next.

I was thinking it could be a drive too, but when I swapped in the 4TB parity drive I got the same errors, and I got them after the 2TB mark during checks and none of my data drives are more than 2TB. Maybe the parity's SATA port on the motherboard is bad?

Maybe the parity's SATA port on the motherboard is bad?

Certainly possible -- but simply to check ... just switch that drive to a different SATA port.  With v5 you can freely change the ports you're using.

Maybe the parity's SATA port on the motherboard is bad?

 

Certainly possible -- but simply to check ... just switch that drive to a different SATA port.  With v5 you can freely change the ports you're using.

Hmm... My board has 6 ports and they're all being used. Can I just swap it with a different drive and be Ok?