Anyone out there an expert on SAS cards?

[DezGalbie]

I previously posted about an issue regarding writing to my HDD - See here for full details of the issue. I managed to fix the drive on that occasion by removing it from my server and scanning/remapping it via an HDD dock on my desktop PC running the Victoria HDD repair program.

But I now have the same issue with another HDD in a very short space of time. I've realised that the common denominator here is a new SAS controller card which both of these drives have been connected to. All of my drives have worked great for some time until I installed this card. This is the SAS card in question.

The question is - Is the SAS card somehow causing this issue? Or is it just a coincidence? Is there anything different I should be doing to stop this happening?

 

[Bishop Crane]

I'm assuming the SAS card is the Array controller?  Or are you just using it as an adaptor for software raid or JBOD up the chain?
Either way the LSI's do have some decent diagnostics built in.
At boot you can normally press Ctrl+C to bring up the adaptor diagnostics (Crtl+A on some cards)
Most of them will have disk and controller diagnostics, worth running some of the basic ones, don't do the drive diagnostics unless you're desperate, they take forever!!
Some of the cards also have a log built in for array errors, worth looking at.

Couple things to check also, does it have a backup battery?  Is it in good/bad health, you can sometimes be better off without the battery if its unhealthy.

Are both drives on the same Port on the card?  If so it could be the port or the Cable, the splitters don't often go, but do sometimes.

Also you say you have 10 drives in your other post, but the SAS card is only an 8 port card or are you using a SAS Expander?  At which point, are both drives connected to the expander?

Last thing, are all the drives from the same batch?  If so you could just be unlucky and it could be a bad batch of drives.

[DezGalbie]

3 minutes ago, Bishop Crane said:

I'm assuming the SAS card is the Array controller?  Or are you just using it as an adaptor for software raid or JBOD up the chain?
Either way the LSI's do have some decent diagnostics built in.
At boot you can normally press Ctrl+C to bring up the adaptor diagnostics (Crtl+A on some cards)
Most of them will have disk and controller diagnostics, worth running some of the basic ones, don't do the drive diagnostics unless you're desperate, they take forever!!
Some of the cards also have a log built in for array errors, worth looking at.

Couple things to check also, does it have a backup battery?  Is it in good/bad health, you can sometimes be better off without the battery if its unhealthy.

Are both drives on the same Port on the card?  If so it could be the port or the Cable, the splitters don't often go, but do sometimes.

Also you say you have 10 drives in your other post, but the SAS card is only an 8 port card or are you using a SAS Expander?  At which point, are both drives connected to the expander?

Last thing, are all the drives from the same batch?  If so you could just be unlucky and it could be a bad batch of drives.

The card is essentially just being used as a means to connect more SATA drives to the system in a JBOD type setup. I have 6 drives connected to the SATA ports on the motherboard, then 4 drives all connected to the same SAS port on the card via a SAS to 4x SATA breakout cable. So there's no SAS expander at play.

No backup battery on the card.

The drives are from different batches but are the same type of Toshiba 3TB drives. They've worked great for a while, and then these 2 drives have issues shortly after installing this card. The timing is what makes me suspect the card, but I suppose it could be the cable too since that's new as well.

I'll have a look around at the adaptor diagnostics tomorrow on my day off. There may be something in there.

Do you have any suggestions to narrow it down and determine which part(s) are at fault?

[Bishop Crane]

If the card diagnostics don't show up anything, and both failed drives are connected to the SAS card that does raise suspicion against it.
Another thing to try to be certain is if you swap the drive to an onboard adaptor does it work ok?  Or does it at least possibly give you SMART errors?

If the SAS card is at fault its notoriously hard to diagnose, if you are only using one of the ports, try the other one.
The 4x cable there isn't much you can do except visually check it for kinks, marks etc.  depending on the card you may be able to do some of the drive diagnostics but like I say that will take a long time.

The one thing I would say is if the link you shared was for the actual card you bought, that is a suspiciously cheap price to pay for a SAS controller.  It would make the 4x cable more expensive than the card itself.  Maybe I'm reading too much into it though!

Also, may be worth looking at the power saving options, to see if it is powering down the drives properly.

One last random thought, its a much older card a PCI 1.0 card is there anything else sharing the same bus?  Its really unlikely that this would cause issues, but it is possible.

[DezGalbie]

2 minutes ago, Bishop Crane said:

If the card diagnostics don't show up anything, and both failed drives are connected to the SAS card that does raise suspicion against it.
Another thing to try to be certain is if you swap the drive to an onboard adaptor does it work ok?  Or does it at least possibly give you SMART errors?

If the SAS card is at fault its notoriously hard to diagnose, if you are only using one of the ports, try the other one.
The 4x cable there isn't much you can do except visually check it for kinks, marks etc.  depending on the card you may be able to do some of the drive diagnostics but like I say that will take a long time.

The one thing I would say is if the link you shared was for the actual card you bought, that is a suspiciously cheap price to pay for a SAS controller.  It would make the 4x cable more expensive than the card itself.  Maybe I'm reading too much into it though!

Also, may be worth looking at the power saving options, to see if it is powering down the drives properly.

One last random thought, its a much older card a PCI 1.0 card is there anything else sharing the same bus?  Its really unlikely that this would cause issues, but it is possible.

Yeah, that's the actual card and buyer that I bought from. It was indeed quite cheap, which only adds to the suspicion! Lol.

I'll have a good look at all this stuff tomorrow - the diagnostics, power saving etc. Hopefully I will find something there.

If I don't find anything in there then I'll try switching the drives around to see what difference that makes. The problem always seems to crop up when writing data. I haven't had any problems simply reading the data (yet). So maybe if I connected 4 drives which are already full of backed-up data to the card, and have the other drives which are still being written to connected via the motherboard SATA ports then things will run smoothly? (In the short term anyway until I can try a better/more reassuringly costly SAS controller)

[Bishop Crane]

Ah, ok, if your issues is specifically when writing, and not reading, then it would suggest an issue with the way the write cache is working.

If you're just using it as a pass through and totally fine accepting the normal power outage risks then disable the write behind feature, I can't remember if this is a global or per logical drive setting.

 

I think on LSI its called Write Back, and when you disable it you put it in Write Through mode.