-
Important Announcement for the TrueNAS Community.
The TrueNAS Community has now been moved. This forum has become READ-ONLY for historical purposes. Please feel free to join us on the new TrueNAS Community Forums
SAS disks
- Thread starter Octopuss
- Start date
- Joined
- Feb 15, 2014
- Messages
- 20,194
Depends on the server. You need a chassis with an appropriate backplane, it's not very safe to do this with bare cables. Although, unplugging a running disk shouldn't be too bad. Plugging in a disk is fairly likely to cause trouble with the PSU.
rvassar
Guru
- Joined
- May 2, 2018
- Messages
- 972
The problem with bare cables you have no mechanical alignment, no rails to keep everything straight as the insertion occurs. So you run the risk of bending pins, and shorting PSU supply rails. It's just not worth the risk.
I am glad I found this thread as I am in the same boat as @Octopuss in that I was looking at used SAS drives on Ebay to use as a pool.
@Octopuss, which seller was willing to provide SMART info to you prior to purchase? I have asked 7 different SAS drive sellers and they all responded in the negative.
Lots of interesting info related to block size etc that I might have to go over as well since the sellers I found that sell a mix of 512 and 520 block size randomly. As long as the formatting can be changed to 512 then it should be ok to run together, correct?
The sellers that I have been looking at, for 3TB sizes, have the HGST HUS723030ALS640 & Seagate ST3000NM0023 drives. Should either of those model be of concern?
I am not looking for speed like the OP. The reason I was looking at used SAS drives was
@Octopuss, which seller was willing to provide SMART info to you prior to purchase? I have asked 7 different SAS drive sellers and they all responded in the negative.
Lots of interesting info related to block size etc that I might have to go over as well since the sellers I found that sell a mix of 512 and 520 block size randomly. As long as the formatting can be changed to 512 then it should be ok to run together, correct?
The sellers that I have been looking at, for 3TB sizes, have the HGST HUS723030ALS640 & Seagate ST3000NM0023 drives. Should either of those model be of concern?
I am not looking for speed like the OP. The reason I was looking at used SAS drives was
- I wanted to create a ZFS pool on proxmox as a backup to my main TrueNAS pool
- I was planning on a 6 drive RAIDZ2 pool and keep 1 HDD as cold spare -- so I would be buying a total of 7 drives.
- at $15-$25 a pop, they are significantly cheaper than new SATA drives which cost around $70 for WD Red Plus 3TB and $105 for Seagate Ironwolf 3TB
I didn't get any SMART results from the seller, but there was a screenshot from HD Sentinel which I also use, so it was good enough for me.
https://www.ebay.com/usr/synergyindustrial is the one I bought the disks from. He was fairly willing to communicate too.
During the course of looking for suitable disks, I figured it was probably best to avoid anything older than five years or so, or stuff that had too many hours on record.
I haven't started using the disks yet because I ran into all kinds of complications, but at least I know for a fact they are physically without errors.
https://www.ebay.com/usr/synergyindustrial is the one I bought the disks from. He was fairly willing to communicate too.
During the course of looking for suitable disks, I figured it was probably best to avoid anything older than five years or so, or stuff that had too many hours on record.
I haven't started using the disks yet because I ran into all kinds of complications, but at least I know for a fact they are physically without errors.
- Joined
- Mar 18, 2014
- Messages
- 1,177
From anecdotal evidence the HGST drives are more reliable than Seagate.
Also try for 7K4000 generation drives : HUS724XXXXX
rvassar
Guru
- Joined
- May 2, 2018
- Messages
- 972
It doesn't have to be anecdotal...
Backblaze Drive Stats for Q3 2021
By September 30, 2021, Backblaze was managing a whopping 194,749 drives. Here’s a deeper look into the SSD and HDD drive stats for Q3 2021 at Backblaze.
I bought my first set of 5 drives for $13 bucks a pop. Cheap enough to not fret too much if the HDDs fail faster than expected. Will see what condition they are in when they arrive. Getting ready to put them in my server and run badblocks/SMART etc.
I will test these out first and then buy another 2-- so that I can create a RAIDz2 pool with 6 disks and keep 1 spare -- assuming all drives don't fail during testing.
I will test these out first and then buy another 2-- so that I can create a RAIDz2 pool with 6 disks and keep 1 spare -- assuming all drives don't fail during testing.
- Joined
- Oct 23, 2020
- Messages
- 1,919
Certainly too hot for my taste. If you look at the Backblaze paper, anything over 30 C has a significant impact, when the drives get 3-4 years old. Also, I remember that I recently read in the fine print of a data sheet (Toshiba?) that the specified MTBF failure etc. were valid only below 30 C. I would encourage you to do some more research on this and form your own opinion.
I have been running badblocks on my 5 drives and it has taken 24 hours to finish 77% of the very first test. Seems it will take about 32 hours for each pass. I have run badblocks on enough drives to know that there are 4 tests each with a Writing phase and a Reading & Comparing phase each. I know all phases are not linear but assuming it takes 32 hours for each pass, it would take 256 hours = almost 11 days to finish. :(
Not to mention the 6 hours it took for the smartctl long test .. and a 2nd round of smartctl long would add another 6 or so hours in order for the entire burn-in process to complete.
I wish it was faster.
EDIT: forgot to mention I got the Seagate ST3000NM0023 drives -- all 5 manufactured in 2015 with between 45065-45073 hours each. So just a shade over 5 years of power on hours. Start/Stop cycles were between 100-109
Not to mention the 6 hours it took for the smartctl long test .. and a 2nd round of smartctl long would add another 6 or so hours in order for the entire burn-in process to complete.
I wish it was faster.
EDIT: forgot to mention I got the Seagate ST3000NM0023 drives -- all 5 manufactured in 2015 with between 45065-45073 hours each. So just a shade over 5 years of power on hours. Start/Stop cycles were between 100-109
Last edited:
Bought another set of 2 HGST HUS724030ALS640 3TB drives to complete the 6xRAIDZ2 + 1cold spare. These were manufactured in Dec 2015 .
These HGST didn't show up in fdisk -l at all and couldn't even connect to the drive but gdisk could connect to the drive. Looked up smartctl info and sure enough they were 520 formatted and the best thing -- both drives only had 19 hours of power-on time. I didn't know that fdisk couldn't connect to anything but 512 formatted drives.
In any case, I ran sg3_utils to reformat it to 512. Now running badblocks on both which will probably take about 72 hours per drive as that's how long it took for my Seagate drives when I ran with block size 4092 and -c flag set as 256. Running them both in tmux... so I should be able to set up the RAIDZ2 in about 3-4 days.
These HGST didn't show up in fdisk -l at all and couldn't even connect to the drive but gdisk could connect to the drive. Looked up smartctl info and sure enough they were 520 formatted and the best thing -- both drives only had 19 hours of power-on time. I didn't know that fdisk couldn't connect to anything but 512 formatted drives.
In any case, I ran sg3_utils to reformat it to 512. Now running badblocks on both which will probably take about 72 hours per drive as that's how long it took for my Seagate drives when I ran with block size 4092 and -c flag set as 256. Running them both in tmux... so I should be able to set up the RAIDZ2 in about 3-4 days.
