Does mkswap -c secure?

FeyFre · 01-16-2012, 10:31 AM

Hi

Recently I got into possession the one moderately old hard disk. So decided to use it partially as swap device. Created 4GB size partition at the beginning of disk and run mkswawp -c /dev/sdb1.

Question:
mkswap did its work and notified me about a number of bad blocks. Is is safe to use such partition as swap? I.e. can I hope those bad blocks became marked as such and will never be used by kernel?

Regards, Yaroslav

ponce · 01-16-2012, 10:42 AM

no, it's not safe: for what I know (the man page), mkswap doesn't mark bad blocks, it just reports them.

bad blocks under a certain number should be handled at SMART level and the operating system should not even see them: if you are seeing them it means that the drive is no more reliable.

read the old mkswap man page:

Quote:

Nobody wants to use a swap space with hundreds of bad blocks.
(I would not even use a swap space with 1 bad block.)

FeyFre · 01-16-2012, 11:33 AM

I have read somewhere too about mkswap behaviour with bad-blocks(about not marking, but counting only), but cannot remember where(and definitely not in man mkswap), so decided to try...

Quote:

bad blocks under a certain number should be handled at SMART

SMART said device has 96(or something like) relocated blocks. I don't know how much it for 100GB HDD.
In any case I have created ext2 partition(via mkfs.ext2 -c /dev/sdb2, mkfs.ext2 -cc takes more then 8 hours to finish, and I never let it done) on the rest of disk, and it made copy of inode tables on bad block(fsck reports the same write problem on the same block each time I run it). Does it means mkfs.ext2 doesn't marks bad blocks either?

PS: now I'm sure, why my system freezed once

Neither remote nor console access was available, but animation in flashplayer continued to play independently(onboard shader?).

titopoquito · 01-16-2012, 12:06 PM

Quote:

Originally Posted by FeyFre

SMART said device has 96(or something like) relocated blocks. I don't know how much it for 100GB HDD.

Check to make sure the "raw value" (last column) of the SMART output in that line. I am no expert on smartctl and cannot read VALUE, WORST and THRESH values to make any meaning. From my experience, if you have one bad block you will get more and more bad blocks very fast. Show me one bad block and I am running to get a new harddisk. So to answer your question - yes, I think that 96 bad blocks (if that's the "RAW_VALUE" output) is way too much.

FeyFre · 01-16-2012, 02:18 PM

Code:

SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x000f   100   098   051    Pre-fail  Always       -       1
  3 Spin_Up_Time            0x0007   100   100   025    Pre-fail  Always       -       3456
  4 Start_Stop_Count        0x0032   094   094   000    Old_age   Always       -       6521
  5 Reallocated_Sector_Ct   0x0033   096   096   011    Pre-fail  Always       -       9
  7 Seek_Error_Rate         0x000f   100   100   051    Pre-fail  Always       -       0
  8 Seek_Time_Performance   0x0025   100   100   015    Pre-fail  Offline      -       9497
  9 Power_On_Half_Minutes   0x0032   099   099   000    Old_age   Always       -       7393h+28m
 10 Spin_Retry_Count        0x0033   100   100   051    Pre-fail  Always       -       1
 11 Calibration_Retry_Count 0x0012   100   100   000    Old_age   Always       -       0
 12 Power_Cycle_Count       0x0032   097   097   000    Old_age   Always       -       3596
194 Temperature_Celsius     0x0022   145   094   000    Old_age   Always       -       31
195 Hardware_ECC_Recovered  0x001a   100   100   000    Old_age   Always       -       1531
196 Reallocated_Event_Count 0x0032   086   086   000    Old_age   Always       -       32
197 Current_Pending_Sector  0x0012   100   100   000    Old_age   Always       -       0
198 Total_Offl_Uncorrectabl 0x0030   100   100   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x003e   200   200   000    Old_age   Always       -       0
200 Multi_Zone_Error_Rate   0x000a   100   100   051    Old_age   Always       -       0
201 Soft_Read_Error_Rate    0x000a   100   100   051    Old_age   Always       -       0

And I beginning to do not trust SMART. I know this HDD is quite old, but my primary HDD is no more than half year aged and it(SMART) shows very bad statistics(worst than for this HDD), but I have not any problems with it.

55020 · 01-16-2012, 02:19 PM

AFAIK bad blocks only get revectored when the block is *written*. Neither mkswap -c or mke2fs -c perform writes, so any bad blocks will stay bad. (But mke2fs -c -c does do writes, see man mke2fs).

Look at 'smartctl -a /dev/sdb' to see how long ago all those errors happened. If they all happened at the same time a long time ago, the errors may have been caused by an external event (low external temperature, or being dropped, or power brown-out) and the disk *may* be ok.

To clear the bad blocks, I would do 'dd if=/dev/zero of=/dev/sdb bs=16384', and then do 'dd if=/dev/sdb of=/dev/null bs=16384', and then look at dmesg. If there were no kernel errors logged in the second 'dd', the disk *may* be ok.

But I certainly wouldn't put the only copies of photos of my children on that disk...

titopoquito · 01-16-2012, 04:52 PM

Quote:

Originally Posted by FeyFre

Code:

ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  5 Reallocated_Sector_Ct   0x0033   096   096   011    Pre-fail  Always       -       9

And I beginning to do not trust SMART. I know this HDD is quite old, but my primary HDD is no more than half year aged and it(SMART) shows very bad statistics(worst than for this HDD), but I have not any problems with it.

As I understand it you have 9 reallocated sectors. Do what 55020 suggested and make a backup of your important files. Did you already perform a long disk self test with smartctl? Might be a bit paranoid, but in my case when the first reallocated sectors occured on any disk others came along very quickly. I wouldn't be too worried about these other "pre-fail"s though.

FeyFre · 01-16-2012, 06:14 PM

Quote:

a backup of your important files.

Using that disk as a storage of downloadable, rarely used data(can be redownloaded in a few hours).

onebuck · 01-17-2012, 10:29 AM

Hi,

I would get the manufactures disk diagnostics and perform to hopefully get a reliable result. You should not use the drive until certain that the bad blocks are indeed 'bad'. Especially within areas that are to be used for swap.

Most certainly do not use with a swap partition on it. Not reliable and you will just open Pandora's box to use a swap partition with bad blocks. You could use a 'swapfile' by 'dd if=/dev/zero of=swapfile bs=1024 count=65536' then use mkswap on that 'swapfile' to confirm and create the swapfile with no errors. Do a 'man mkswap' to fully understand.
HTH!

FeyFre · 01-17-2012, 10:46 AM

Quote:

You could use a 'swapfile' by 'dd if=/dev/zero of=swapfile bs=1024 count=65536'

Yse, I do know about 'swapfile'. Really, I don't need swap at all(it is not so easy to fill up 2GB RAM). Primary goal of using swap is to have ability to go into S4. Apparently swapfile is not usable solution here yet(at least I do not know any case).