hi guys

 

i have a 14 disc array of mainly 2.5 ssd's.  array running for years.

last week noticed 1 disc with red cross (disc 8), ordered spares.

 

during rebuild #1 i had errors on 1 disc (disc 12), like 300.000 - but data sync success

then swapped disc #12 & rebuilt - success, array returned to normal

 

but my directories are all gone in my mapped drive.

 

if i see contents of a random disc, i do see contents there (directories like "music", 'audiobooks", "movies" etc)

 

this a 30 tb array so i cant quickly swap all that data around on a temporary solution.

 

how do i restore my original shares?

 

there is one error in log: kernel: XFS (md8p1): metadata I/O error in "xfs_imap_to_bp+0x50/0x70 [xfs]" at daddr 0x37e3f730 len 32 error 117

 

thanks in advance to all your help

 

steven

Edited November 18, 20241 yr by steven_76

dia added

cube-diagnostics-20241118-2114.zip

If there were read errors on another device during a rebuild, and with only single parity, the rebuilt disk will likely have some corruption, you should have stopped it.

 

Check filesystem on disk8, run it without -n.

 

Also seeing some ATA errors for multiple disks, you should investigate that.

13 hours ago, JorgeB said:

If there were read errors on another device during a rebuild, and with only single parity, the rebuilt disk will likely have some corruption, you should have stopped it.

 

Check filesystem on disk8, run it without -n.

 

Also seeing some ATA errors for multiple disks, you should investigate that.

i owe you a beer 😀 the check on disc 8 took 1sec and i immediately solved my problem. seems i have no data loss, but share is online again and mapped, very happy with this.

 

about ATA errors, which lines in config dump should i be looking for to ID those disc numbers?

im gradually swapping out all spinners for SSD, as finances allow. i might be wrong but i take it those are less "wear & tear" sensitive

beer is no longer owed, cheers buddy

11 hours ago, steven_76 said:

about ATA errors, which lines in config dump should i be looking for to ID those disc numbers?

Look at the lsscsi.txt file in the diags, for example, there are issues with ata17:

 

Nov 18 07:50:07 CUBE kernel: ata17.00: configured for UDMA/33
Nov 18 07:54:47 CUBE kernel: ata17: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
Nov 18 07:54:53 CUBE kernel: ata17.00: qc timeout after 5000 msecs (cmd 0xec)
Nov 18 07:54:53 CUBE kernel: ata17.00: failed to IDENTIFY (I/O error, err_mask=0x4)
Nov 18 07:54:53 CUBE kernel: ata17.00: revalidation failed (errno=-5)
Nov 18 07:54:53 CUBE kernel: ata17: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
Nov 18 07:54:53 CUBE kernel: ata17.00: supports DRM functions and may not be fully accessible

 

And in lscsi.txt you can see that ata17 is sdq:

 

[10:0:0:0]   disk    ATA      CT1000MX500SSD1  033   /dev/sdq   /dev/sg16
  state=running queue_depth=32 scsi_level=6 type=0 device_blocked=0 timeout=30
  dir: /sys/bus/scsi/devices/10:0:0:0  [/sys/devices/pci0000:00/0000:00:15.0/0000:05:00.0/ata17/host10/target10:0:0/10:0:0:0]

 

 

Start by replacing the cables for the affecte devices and monitor for new errors.

 

 

 

11 hours ago, steven_76 said:

beer is no longer owed, cheers buddy

Thanks!