NAS config verification

[cyberjock]

The LSI 9201-16i definitely does pass SMART functions, because I see them listed for all the other drives. It's just odd that only one drive doesn't list the serial number, which I believe is what's preventing SMART from verifying that drive. Correct me if I'm wrong on that statement.

Not sure.. keep reading.

Heya Mlovelace,

I wasn't able to get the 'smartctl -a /dev/daX' command to work. And just curious, what numbers are you pulling with your LSI board?

So how do you know it passes SMART functions if smartctl -a didn't work. Literally, smartctl -a *is* a query of the SMART data on your drive. So if it's not working then what is working? Getting more confused the more I read.

Every device can be a bottleneck. Might be your SSD, might be your NIC. The only thing you can do is figure out what the least common denominator is and make that faster. That's all.

I don't recommend Plextor SSDs around here. Intel are the way to go for L2ARCs and ZILs. There's plenty of users that have bought other drives (read: cheaper) and they've more often than not found out that the SSD didn't perform as advertised and, in some cases, actually made their server slower because the SSD wasn't fast enough.

All this thread is doing right now is talking about theoretical "what might be the problem". We could easily be talking about the possibility that the universe will end on Dec 31st and be just as productive.

What you need to do is read our stickies, do some of our tests for finding bottlenecks and use those results to figure out where your limitations are. Until then we're spinning wheels and getting nowhere fast. ;)

 

[Mlovelace]

*double post*

 

[Mlovelace]

Not sure.. keep reading.


So how do you know it passes SMART functions if smartctl -a didn't work. Literally, smartctl -a *is* a query of the SMART data on your drive. So if it's not working then what is working? Getting more confused the more I read.

The HBA passes smart for sure I have 4 of the 9201-16i's in service and I can query smart on all devices...

Another question, are you substituting /dev/'daX' in the command for your actual drive number ig. da0, da1, da2??

 

[sheld0r]

So how do you know it passes SMART functions if smartctl -a didn't work. Literally, smartctl -a *is* a query of the SMART data on your drive. So if it's not working then what is working? Getting more confused the more I read.

At the console it states, 'FAILED SMART self-check for device /dev/da4. So I assumed that SMART was running for all the disks, but I should never assume and always verify.

Another question, are you substituting /dev/'daX' in the command for your actual drive number ig. da0, da1, da2??

I had an extra space entered, my mistake.

Results:
SMART overall-health self-assessment test result: FAILED!

So the only way to find which disk has failed, is to match the slot with the SN, correct? Because FreeNAS doesn't know which enclosure is used in the various setups. Is that accurate?

@cyberjock
I completely agree and understand your statement. I was more so expressing my thoughts on the forums, I don't mean to waste anybody's time. I realize I have some reading to do and testing. I'll dig deeper on my end and post up my findings later.

 

[Mlovelace]

Results:
SMART overall-health self-assessment test result: FAILED!

You sir have a bad drive. Follow the steps in the manual to replace the drive. Since you don't know which bay it's in, after you offline the drive, you can power down and sort it out by s/n; power back on and finish the replace.

 

[sheld0r]

I guess when you purchase 54 drives, you're bound to get a bad one :(

Thanks for the command, that's definitely one to keep handy!

 

[superwad]

54TB of drives, not 54 drives right?

I saw a nice piece of advice in the forums today. In the description field for the list of drives, stick in the physical location in your case that the particular drive is in. Then when a drive fails, you look at the location and match it to the serial number instead of looking at each drive for the correct serial.

 

[sheld0r]

Nope, 54 drives! I built 3 arrays.

Good advice! I guess I'm spoiled from my SANs with the red LED of the drive that failed. lol

 

[superwad]

Wow, that's a lot of drives. Looks like your patience to build paid off :D What kind of size are you talking about after everything is all said and done? I started mine up with 18x6TB, in 3x6-drive vdevs. I'm thinking about dropping in another 6x3TB to finish up the case.

What kind of case are you using? I have a Supermicro 4U and as far as I know the backplane should still light up when a drive physically fails. Not sure about if SMART fails or something else on the software layer that FreeNAS cares about though. I tried pulling out a drive live and let me say, you don't want to do that. Follow the advice on offlining a drive before you pull one even for testing.

 

[sheld0r]

I really wanted to use a 4U Supermicro case! But I built these arrays as cheap as possible, while still sneaking in some performance. I used the Norco RPC-4020 chassis. Two of them are running Debian 7 w/ 28.6TB useable space and the FreeNAS provides 31.9TB of useable space. I've also created the arrays with 3x6 drive for the Debian boxes, and attempted to create one large vdev on the FreeNAS, but learned that's not a good idea.

You dropped in 6TB drives! Very nice!! Which ones? I'm using the RED NAS 5400rpm drives. I'm a little disappointed, as they now released the PRO series of the RED NAS drives and those are 7200rpm.

 

[cyberjock]

54TB of drives, not 54 drives right?

I saw a nice piece of advice in the forums today. In the description field for the list of drives, stick in the physical location in your case that the particular drive is in. Then when a drive fails, you look at the location and match it to the serial number instead of looking at each drive for the correct serial.

That doesn't always work. In fact, in some situations it is entirely possible for you to put info on the description field and then for it to be incorrect later. The only way you should label it is by labeling the trays themselves.

 

[superwad]

I really wanted to use a 4U Supermicro case! But I built these arrays as cheap as possible, while still sneaking in some performance. I used the Norco RPC-4020 chassis. Two of them are running Debian 7 w/ 28.6TB useable space and the FreeNAS provides 31.9TB of useable space. I've also created the arrays with 3x6 drive for the Debian boxes, and attempted to create one large vdev on the FreeNAS, but learned that's not a good idea.

You dropped in 6TB drives! Very nice!! Which ones? I'm using the RED NAS 5400rpm drives. I'm a little disappointed, as they now released the PRO series of the RED NAS drives and those are 7200rpm.

I got the WD RED WD60EFRX, so they're probably the 5400rpm as they're certainly not pro drives.

That doesn't always work. In fact, in some situations it is entirely possible for you to put info on the description field and then for it to be incorrect later. The only way you should label it is by labeling the trays themselves.

Fair point. I have the location of all my drives in the trays in a spreadsheet. I just need to go through the trouble of making little stickers now :)