How to reformat HDD & SSD to 512B Sector Size

[Sharin]

I executed the command "sg_format -v --format --size=512 da0" on a Huawei storage TOSHIBA AL14SEB090N 520-format 900G SAS hard drive, and after a long wait, I successfully obtained a 512-format usable hard drive. （DELL R730xd with a HBA330）
Although this operation successfully made the 520 hard drive readable and writable, the actual speed is halved.
A normal 10k 900G drive has a throughput of about 200MB/s using dd if=zero of=/dev/da0 bs=1m, while this drive, reformatted using sg_format, shows only 95MB/s in actual performance.
cannot accept a 50% performance loss, so in the end, I could only change it back to 520 format and plan to sell it.

root@mfsbsd:~ # sg_format -v --format --size=520 da0 TOSHIBA AL14SEB090N 080C peripheral_type: disk [0x0] PROTECT=0 Unit serial number: X870A24SF4PD LU name: 50000398f82083e1 mode sense(10) cdb: [5a 00 01 00 00 00 00 00 fc 00] Mode Sense (block descriptor) data, prior to changes: Number of blocks=1758174768 [0x68cb9e30] Block size=512 [0x200] mode select(10) cdb: [55 11 00 00 00 00 00 00 1c 00] A FORMAT UNIT command will commence in 15 seconds ALL data on da0 will be DESTROYED Press control-C to abort A FORMAT UNIT command will commence in 10 seconds ALL data on da0 will be DESTROYED Press control-C to abort A FORMAT UNIT command will commence in 5 seconds ALL data on da0 will be DESTROYED Press control-C to abort Format unit cdb: [04 18 00 00 00 00] Format unit command launched without error Format unit has started Format unit in progress, 4.50% done Format unit in progress, 5.09% done ... ...

 

[napalm_atx]

A normal 10k 900G drive has a throughput of about 200MB/s using dd if=zero of=/dev/da0 bs=1m, while this drive, reformatted using sg_format, shows only 95MB/s in actual performance.

Check mode pages settings with sd_parm

 

[Bilbobagg99]

If you are running windoze at all, goto BPS Converter to obtain a ridiculously easy block converter program that can handle all accesible SAS (and other) drives in your system, converting to 512 so ZFS and other progs can access.

 

[Bilbobagg99]

BTW, i found it would auto recognize all the SAS drives I had online (8) and did the conversion on all of them at the same time. Took about 9 hours for the 2 TB drives I had hooked up. Then I tried it with 8 4 TB drives and it took a little over 12 hours.

 

[camelshoko]

Try sg_format with the -S (secure erase mode) flag. May work better.

Can someone try to solve this riddle for me?

I have four HUSMR1610ASS204s that presumably came out of a 3PAR. They report themselves as being an HP-specific part number, HSCP0920S5xnNMRI, unlike the HUSMR1010ASS204 model units (same HP P/N) I have from the same seller. When I did my first sg_format -F -6 /dev/sgX to them, they mysteriously lost 70GB of capacity/blocks???

Screenshot below contains 3 drives, a 1010 and two 1610s. the 1010 (sg2) happily reformatted from 1015GB of 520b sectors to 1000GB of 512b, with the block count remaining unchanged. As you can see it has the same number of blocks as sg4, which is an un-sg_formatted 1610.

But sg3, the 1610 which I'd already formatted to 512b, has gone from 1,953,525,168 blocks to 1,788,974,512 blocks - 164,550,656 blocks have disappeared entirely! sg4 did this also after I formatted it;

What the heck is going on here? Smartmontools reports 0% write endurance used and <35TB lifetime writes - nothing for a drive of this size and rating. Even reformatting back to 520 bytes (which I did before the screenshot above) hasn't given me back the 70GB, nor has an sg_sanitize or running sg_format with the -S (secure erase) flag.

Anyone have a clue what's going on here?

Edit: quick update to this, took me less time to fix than to make the original post... oops. If I run an sg_format -r --count=1953525168 /dev/sgX (command the device to resize itself) after changing the block size, the capacity comes back and the drive seems to be totally OK with it;

I suspect this has to do with the HP firmware and the fact that these are labelled as "920GB" when they're actually 1000GB units internally (HUSMR1010ASS20 / HUSMR1610ASS20 parts are 1000GB parts) - the HP firmware appears to recommend a 930GB capacity instead of 1000, even though they were reporting 1000GB of capacity out of the box.

Weird.
Anyway, hopefully that helps someone who bought one of these weird 1610s.

Hi, Dear

Could you tell me why the count is 1953525168
How to calculate the value.

Thanks.

 

[shpitz461]

Could you tell me why the count is 1953525168

1953525168 is number of blocks in total for the whole drive, just below it is the block size: 512.

1953525168 x 512 = 1000204886016, which means it's a 1TB drive (931GiB).

 

[camelshoko]

1953525168 is number of blocks in total for the whole drive, just below it is the block size: 512.

1953525168 x 512 = 1000204886016, which means it's a 1TB drive (931GiB).

Thanks very much!

 

[Dell-icious]

http://imgur.com/a/Q2U9sVX

Has anyone else had this problem? I have a batch of 8TB hard drives and it won't let me format them. I'm using Windows 10 with an HBA card. D:

 

[Behemot]

Hi, I'll join the thread even though this is likely not problem of formatting right away. Just obtained some SM1633s from NetApp. Hooked them to me good ol' HP branded, long time ago original-LSI-FW-reflashed 1064E HBA, but no joy. So this is not problem with drive formatting, it does not even see anything on the port at all. Other drives work fine (mostly 15k HDDs and also two Seagate 1200 units which came together with the SM1633s, just poring me game installations on them ATM).

So I wonder whether it has to do with the HBA being too old, only SAS 3 Gb/s, that they are not detected whatsoever? Maybe these drives with their (possibly) custom FW do not even communicate over SAS-1? I looked around what I can get, also because of the fact that while these 1064E state they communicate over PCIe 1.0 ×8 in their DS (as well as what AIDA64 state, so the chipset itself believes it is 8 PCIe linney), the cards interface itself is just 4 PCIe lines long.

And I get only some 200 MB/s even from these Seagate 1200 SSDs, which is laughable, according to HD Tune. I've expected speeds closer to 300 MB/s, the lines maximum. I do get as much as 550 MB/s read with two of them stripped using volume-oriented benchmark (CrystalDiskMark), but less than 200 MB/s write. Some LSI 9300-i8 of fleeabay caught my eye as they sell dirty cheap, incl. the cabling and with supposedly IT mode (HBA) firmware. And there seems to be drivers available for Win 7. Could I expect everything to run smoother with this controller? Possibly also the SM1633s to get recognised?