13

I have a software RAID 1 array on RHEL. I am getting this error emailed to me each morning: Device: /dev/sda [SAT], 1 Currently unreadable (pending) sectors

When I run a test on sda (or sdb) everything appears to pass. Am I missing something?


mdstat shows the RAID array is fine:

# cat /proc/mdstat 
Personalities : [raid1] 
md5 : active raid1 sdb5[1] sda5[0]
      108026816 blocks [2/2] [UU]

md1 : active raid1 sdb1[1] sda1[0]
      511936 blocks [2/2] [UU]

md2 : active raid1 sda2[0] sdb2[1]
      805306304 blocks [2/2] [UU]

md3 : active raid1 sda3[0] sdb3[1]
      62914496 blocks [2/2] [UU]

unused devices: <none>

Here is the output of: # smartctl -q noserial -a /dev/sda

I have also tried running: # smartctl -t long /dev/sda


smartctl 5.43 2012-06-30 r3573 [x86_64-linux-2.6.32-279.9.1.el6.x86_64] (local build)
Copyright (C) 2002-12 by Bruce Allen, http://smartmontools.sourceforge.net

=== START OF INFORMATION SECTION ===
Model Family:     Hitachi Ultrastar A7K1000
Device Model:     Hitachi HUA721010KLA330
Firmware Version: GKAOAB0A
User Capacity:    1,000,204,886,016 bytes [1.00 TB]
Sector Size:      512 bytes logical/physical
Device is:        In smartctl database [for details use: -P show]
ATA Version is:   7
ATA Standard is:  ATA/ATAPI-7 T13 1532D revision 1
Local Time is:    Sun May 21 17:51:42 2017 CDT
SMART support is: Available - device has SMART capability.
SMART support is: Enabled

=== START OF READ SMART DATA SECTION ===
SMART overall-health self-assessment test result: PASSED

General SMART Values:
Offline data collection status:  (0x84) Offline data collection activity
                    was suspended by an interrupting command from host.
                    Auto Offline Data Collection: Enabled.
Self-test execution status:      (   0) The previous self-test routine completed
                    without error or no self-test has ever 
                    been run.
Total time to complete Offline 
data collection:        (15354) seconds.
Offline data collection
capabilities:            (0x5b) SMART execute Offline immediate.
                    Auto Offline data collection on/off support.
                    Suspend Offline collection upon new
                    command.
                    Offline surface scan supported.
                    Self-test supported.
                    No Conveyance Self-test supported.
                    Selective Self-test supported.
SMART capabilities:            (0x0003) Saves SMART data before entering
                    power-saving mode.
                    Supports SMART auto save timer.
Error logging capability:        (0x01) Error logging supported.
                    General Purpose Logging supported.
Short self-test routine 
recommended polling time:    (   1) minutes.
Extended self-test routine
recommended polling time:    ( 255) minutes.
SCT capabilities:          (0x003f) SCT Status supported.
                    SCT Error Recovery Control supported.
                    SCT Feature Control supported.
                    SCT Data Table supported.

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     0x000b   098   098   016    Pre-fail  Always       -       4
  2 Throughput_Performance  0x0005   100   100   054    Pre-fail  Offline      -       0
  3 Spin_Up_Time            0x0007   122   122   024    Pre-fail  Always       -       550 (Average 591)
  4 Start_Stop_Count        0x0012   100   100   000    Old_age   Always       -       68
  5 Reallocated_Sector_Ct   0x0033   100   100   005    Pre-fail  Always       -       0
  7 Seek_Error_Rate         0x000b   100   100   067    Pre-fail  Always       -       0
  8 Seek_Time_Performance   0x0005   100   100   020    Pre-fail  Offline      -       0
  9 Power_On_Hours          0x0012   094   094   000    Old_age   Always       -       43202
 10 Spin_Retry_Count        0x0013   100   100   060    Pre-fail  Always       -       0
 12 Power_Cycle_Count       0x0032   100   100   000    Old_age   Always       -       68
192 Power-Off_Retract_Count 0x0032   100   100   000    Old_age   Always       -       751
193 Load_Cycle_Count        0x0012   100   100   000    Old_age   Always       -       751
194 Temperature_Celsius     0x0002   090   090   000    Old_age   Always       -       66 (Min/Max 17/72)
196 Reallocated_Event_Count 0x0032   100   100   000    Old_age   Always       -       0
197 Current_Pending_Sector  0x0022   100   100   000    Old_age   Always       -       1
198 Offline_Uncorrectable   0x0008   100   100   000    Old_age   Offline      -       0
199 UDMA_CRC_Error_Count    0x000a   200   200   000    Old_age   Always       -       0

SMART Error Log Version: 1
No Errors Logged

SMART Self-test log structure revision number 1
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Short offline       Completed without error       00%     43186         -
# 2  Extended offline    Completed without error       00%     43170         -
# 3  Short offline       Completed without error       00%     43162         -
# 4  Short offline       Completed without error       00%     43138         -
# 5  Short offline       Completed without error       00%     43114         -
# 6  Short offline       Completed without error       00%     43090         -
# 7  Short offline       Completed without error       00%     43066         -
# 8  Short offline       Completed without error       00%     43042         -
# 9  Extended offline    Completed without error       00%     43031         -
#10  Short offline       Completed without error       00%     43024         -
#11  Short offline       Completed without error       00%     43018         -
#12  Extended offline    Completed without error       00%     43002         -
#13  Short offline       Completed without error       00%     42994         -
#14  Short offline       Completed without error       00%     42970         -
#15  Short offline       Completed without error       00%     42946         -
#16  Short offline       Completed without error       00%     42922         -
#17  Short offline       Completed without error       00%     42898         -
#18  Short offline       Completed without error       00%     42874         -
#19  Short offline       Completed without error       00%     42850         -
#20  Extended offline    Completed without error       00%     42833         -
#21  Short offline       Completed without error       00%     42826         -

SMART Selective self-test log data structure revision number 1
 SPAN  MIN_LBA  MAX_LBA  CURRENT_TEST_STATUS
    1        0        0  Not_testing
    2        0        0  Not_testing
    3        0        0  Not_testing
    4        0        0  Not_testing
    5        0        0  Not_testing
Selective self-test flags (0x0):
  After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.
10

Your SMART Current Pending Sector has value 1. This means that there is a bad sector on the disk and the drive firmware is unable to reallocate it, but you still have zero reallocated sectors count so it probably recoverable even though your drive have been running for 5 years in not very healthy environment - temperatures are up to 72 C°.

You can try finding this bad sector with dd if=/dev/sda of=/dev/null and then you'll need to remap this sector by overwriting it.

7
  • Can I run dd if=/dev/sda of=/dev/null on an active machine or do I need to run that off of a live disc, etc.? – MSF004 May 22 '17 at 12:44
  • If your disk subsystem isn't very heavily loaded you can ran dd safely - it will make just a bit of performance impact – AlexD May 22 '17 at 13:52
  • 2
    badblocks is more advanced utility and you will need it if you unable to remap bad block on firmware level but some of its modes are destructive, so you need to be careful when using it. I would replace such aging drive on the first error without bothering myself with recovery but replacing also carries some risks especially if it is done by some overworked technician at a datacenter. It is up to you to weight all risks and make decision. – AlexD May 22 '17 at 15:01
  • 1
    @MSF004 badblocks command works on filesystem level. That is why you need run it on /dev/md2 instead /dev/sda2. You will crash your system if you'll do it on /dev/sda2. – Mikhail Khirgiy May 23 '17 at 8:31
  • 1
    I'm a bit puzzled I thought linux would automatically reallocate sectors, why doesn't it? This seems like precisely the kind of thing that the operating system should do? Also it seems a bit cr*ppy that it doesn't tell you what block either in the mail, or in dmesg. – Owl Jul 29 '19 at 7:30
1

I believe the "197 Current_Pending_Sector" count is is a count of sectors that have failed to read. This does kind of imply that the drive is starting to fail, but it doesn't necessarily mean the drive is bad. If those sectors are re-written the drive firmware can re-map them away and the drive could be fine.

Searching around also finds discussions that suggest some models of SSD drives regularly toggle this to 1 and back, and it might be a mostly harmless firmware bug in the SMART reporting.

So you could just ignore it, and provided your filesystem can handle the occasional bad block read (ie, does some kind of robust raid/redundancy under the hood), it might slowly clear itself as the filesystem overwrites those blocks. If your filesystem cannot handle bad block reads, you could get IO errors every time something tries to read whatever is in that block. You might still be able to recover by finding and deleting that file, and the filesystem will eventually re-write over that sector.

You can also clear the Current_Pending_Sector count by explicitly overwriting those sectors. THIS WILL DESTROY DATA ON THE DISK! This could corrupt the filesystem in ways that could loose all the data on the disk, not just what is lost in the bad sector. So only do this if you can afford to loose all the data on the disk.

You can find the bad sector by running a smart long test;

# smartctl -t long /dev/sda

You can then check the status of the long test to see if its finished, and the LBA of the first error it encountered;

# smartctl -l selftest /dev/sda
smartctl 7.1 2019-12-30 r5022 [x86_64-linux-5.7.0-1-amd64] (local build)
Copyright (C) 2002-19, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF READ SMART DATA SECTION ===
SMART Self-test log structure revision number 1
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Extended offline    Completed without error       00%     17711         12345
# 2  Short offline       Completed without error       00%     17709         -
# 3  Short captive       Interrupted (host reset)      10%       450         -
# 4  Short captive       Interrupted (host reset)      10%       228         -

You can then overwrite the bad sector using dd;

# dd if=/dev/zero of=/dev/sda seek=12345 count=1

Though note that the default bs=512 blocksize might be smaller than the physical sector size on the drive, so you may need to write up to count=8 to wipe it completely. You can then rinse-and repeat the test/overwrite cycle until all the bad sectors are overwritten. Finally, you can check that the count has been zeroed;

$ sudo smartctl -A /dev/sdc | grep Current_Pending_Sector
197 Current_Pending_Sector  0x0012   200   200   000    Old_age   Always       -       0

I experienced this on an old WD 200G HDD that had some kind of hickup that may have been caused by poor hotplug connections that resulted in a Current_Pending_Sector count of 26. When I ran the long test, it passed without finding any bad sectors, but the count was still at 26. I was able to zero the counter it by zeroing out the whole drive with dd;

dd if=/dev/zero of=/dev/sda bs=1M status=progress

Note the bs=1M makes it go much faster, but it's not the 512 sector size so there will be a partial block at the end. Subsequent smartctl long and short tests have all reported the drive as fine.

3
  • In my example case, after using the drive for a while, it started having more Current_Pending_Sector errors, and I believe the drive is actually dying. – Donovan Baarda Jul 24 '20 at 6:33
  • 197 is just the id of that SMART parameter. The number of pending sector errors is 1. – CpnCrunch Sep 2 '20 at 22:02
  • I got an email saying 8 sectors are bad but when I look at the output column LBA_of_first_error, it's all - (0). Does that mean the sectors were already self corrected somehow? – Alexis Wilke Nov 13 '20 at 20:29
0

This just means that you have one bad sector on one drive in your RAID array. There is nothing to worry about at the moment, unless you start getting a lot more bad sectors on that disk. You should NOT try to manually fix the error...that will be done automatically on the first of each month by the raid-check command, which is run from /etc/cron.d/raid-check. You can check that command and run it manually to immediately reallocate the bad sector on the disk:

[root@server]# more /etc/cron.d/raid-check 

0 1 * * Sun root /usr/sbin/raid-check

[root@server]# /usr/sbin/raid-check

This will force mdraid to copy the bad block from the other disk in the RAID array, and it will mark the bad block as unusable.

2
  • I don't see a raid-check anywhere though... – Alexis Wilke Nov 13 '20 at 20:23
  • In Centos it is raid-check, but in other distros it may be called something else. Have a look in /etc/cron.d. – CpnCrunch Nov 15 '20 at 0:52
0

FYI, in my case I got the error about 8 invalid sectors and that did last for a while.

Then I saw this message in my syslog:

Nov 13 06:57:35 monster smartd[3359]: Device: /dev/sdf [SAT], No more Currently unreadable (pending) sectors, warning condition reset after 1 email

So I guess it should fix itself and if not the drive should probably be replaced ASAP.

The first message I can find in my syslog happened on Nov 12 at 06:57:35

Nov 12 06:57:35 monster smartd[3359]: Device: /dev/sdf [SAT], 8 Currently unreadable (pending) sectors

So it looks like it repeated the message until 1 day later and checked the drive again and did not find the errors anymore.

I'm still thinking that I'll get a new hard drive. I can get a 12Tb and have 3 drives in my RAID. So if one dies, I still have 2... It's way more trouble to not have a backup.

My existing drive highest temp. is 48°C. So relatively good. Also the drive ran for only 9599 hours (~400 days) so still very young.

-1

Now You have Current Pending Sector on the sda hard disk. This disk is used in software RAID. There is a start of death. You need to replace this disk.

This disk is designed for desktop PC. It tries to recovery itself (not via RAID). It reads bad sector every several seconds until checksum of bad sector will be good (disk's performance will drop dramatically), then the disk will write the read data to a new reserved sector. But the read data from rescued sector may be wrong often. CRC32 checksum can only indicate first error and it doesn't usable for recovering data (by example - XOR on RAID-5 can used for recovering data). When RAID will have read these data from the bad drive it will give different data, which will be read OK. This situation will produce kernel panic on the system if bad block has important system data. That is why You need to replace bad disk.

3
  • There is 3 incorrect statements in this answer. 1) This disk is specified as enterprise class (and failure rate for enterprise and desktop drives doesn't really differ). 2) The data in each disk sector is protected by ECC and probability of reading incorrect data without it being detected is very low - even with CRC32 you need to read one bad sector 2^32 times to get undetected error (and data should be random on each read) 3) the RAID doesn't read both disks for a single sector and doesn't compare data from both disks so there is no split-brain situation and no kernel panics. – AlexD May 22 '17 at 19:35
  • 1. Enterprise class isn't the same as server class which is used in RAID. RAID disk must die quickly. It doesn't try to survive. 2. The read data from bad sector are random, because data signal level is the same as noise level. 3. Yes. It's my mistake. I have changed my answer. – Mikhail Khirgiy May 23 '17 at 6:35
  • 1) The disk in question supports 'dying fast' (SCT Capabilities line in smartctl output) and rated for RAID usage by its manufacturer 2) The data read from a bad sector usually has just a few bits flipped once. Getting random reading of whole sector from the media which is designed to be static memory is something of very low probability. 3) Modern disks use Reed-Solomon code for error detection and multiple bits correction, CRC32 is just an example for false positive probability. RAID 5 doesn't use parity bits for error detection and relies on the disk hardware and its ECC to detect errors. – AlexD May 23 '17 at 17:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.