Administrace komentářů

Dobry vecer.

Mejme masinu s OS na NVMe SSD a 4x HDD pripojene do SATA radice, pricemz tyto 4x HDD jsou v mdraid5. Nad timto R5 je LUKS, nad LUKS je bezny etx4.

Probehnul pokus dostat celej onboard SATA radic do VM prostrednictvim PCIe passthough... Nedomyslel jsem, ze je potreba upravit startup tak, aby hypervisor pole vubec nedaval dohromady, nebo aby jej hned pred spustenim VM zastavil. Takze vysledek se brzy dostavil. V tento cas startuje VM a zacina probihat passthrough; sda je jiny SATA disk, mimo pole, jehoz odebrani probehlo korektne.

dub 11 08:02:03 gw kernel: sd 0:0:0:0: [sda] Synchronizing SCSI cache
dub 11 08:02:03 gw kernel: sd 0:0:0:0: [sda] Stopping disk
dub 11 08:02:03 gw kernel: ata1.00: disabled
dub 11 08:02:03 gw udisksd[1133]: Unable to resolve /sys/devices/virtual/block/md126/md/dev-sdb/block symlink
dub 11 08:02:03 gw kernel: md/raid:md126: Disk failure on sdb, disabling device.
dub 11 08:02:03 gw kernel: md/raid:md126: Operation continuing on 3 devices.
dub 11 08:02:03 gw kernel: sd 2:0:0:0: [sdb] Synchronizing SCSI cache
dub 11 08:02:03 gw kernel: sd 2:0:0:0: [sdb] Stopping disk
dub 11 08:02:03 gw udisksd[1133]: Unable to resolve /sys/devices/virtual/block/md126/md/dev-sdb/block symlink
dub 11 08:02:03 gw kernel: ata3.00: disabled
dub 11 08:02:03 gw kernel: md/raid:md126: Disk failure on sdc, disabling device.
dub 11 08:02:03 gw kernel: md/raid:md126: Cannot continue operation (2/4 failed).
dub 11 08:02:03 gw kernel: md126: detected capacity change from 23438819328 to 0
dub 11 08:02:03 gw kernel: md/raid:md126: device sde operational as raid disk 3
dub 11 08:02:03 gw kernel: md/raid:md126: device sdd operational as raid disk 0
dub 11 08:02:03 gw kernel: sd 3:0:0:0: [sdc] Synchronizing SCSI cache
dub 11 08:02:03 gw kernel: md/raid:md126: not enough operational devices (2/4 failed)
dub 11 08:02:03 gw kernel: md/raid:md126: failed to run raid set.
dub 11 08:02:03 gw kernel: md: pers->run() failed ...
dub 11 08:02:03 gw kernel: md: md126 stopped.
dub 11 08:02:04 gw kernel: sd 3:0:0:0: [sdc] Stopping disk
dub 11 08:02:04 gw systemd[1]: mdmon@md127.service: Deactivated successfully.
dub 11 08:02:04 gw kernel: md: md126 stopped.
dub 11 08:02:04 gw systemd-udevd[4178]: md126: Process '/usr/bin/unshare -m /usr/bin/snap auto-import --mount=/dev/md126' failed with exit code 1.
dub 11 08:02:04 gw kernel: ata4.00: disabled

  Zbyvajici dva disky se odpojily "normalne", pac md126 uz bylo zastaveny, avsak se zaznamem informace o tom, ze jsou dva disky v haji...  

Proc furt doufam, ze to jeste neni uplne v haji:

1) vsechny disky jsou citelne, NEdoslo k HW chybe, ani k fakticke ztrate redundance

2) nemohlo (?) dojit k poskozeni uzitecnych dat, nakolik pole nebylo ani namountovane, ani LUKS-odemknute

  Takhle vypada stav celeho pole:

mdadm --examine md126
md126:
          Magic : de11de11
        Version : 01.00.00
Controller GUID : ....
 Container GUID : ....
            Seq : 00000015
  Redundant hdr : yes
  Virtual Disks : 2

      VD GUID[0] : 44656C6C:...72
         unit[0] : 0
        state[0] : Failed, Consistent
   init state[0] : Fully Initialised
       access[0] : Read/Write
         Name[0] : data
 Raid Devices[0] : 4 (0@0K 1@0K 2@0K 3@0K)
   Chunk Size[0] : 128 sectors
   Raid Level[0] : RAID5
  Device Size[0] : 3906469888
   Array Size[0] : 11719409664

      VD GUID[1] : 44656C6C:2...1D
         unit[1] : 1
        state[1] : Optimal, Consistent
   init state[1] : Not Initialised
       access[1] : Read/Write
         Name[1] : temp

 Physical Disks : 255
      Number    RefNo      Size       Device      Type/State
         0    b854cc92  3906469888K /dev/sdd        active/Online
         1    21881f31  3906469888K /dev/sdc        active/Offline, Failed
         2    8296e5e6  3906469888K /dev/sdb        active/Offline, Failed, Missing
         3    8d40b56c  3906469888K /dev/sde        active/Online
         4    3e2d301c  2929721344K                 active/Online
         5    46199ca5  2929721344K                 active/Online
         6    fd2d4fe1  2929721344K                 active/Online

 

VD 1 s labelem temp, potazmo fyzickym diskum 4, 5, 6 netreba venovat pozornost, ty patri jinemu poli, ktere neni pritomno (zaloha, offline).  

U examine jednolivych disku nalezicim k VG0 (data) se pak stav lisi podle toho, kdy kterej disk vypadnul. sdb vypadnul jako prvni a pamatuje si stav, kdy bylo vsechno zalite sluncem:

      Number    RefNo      Size       Device      Type/State
         0    b854cc92  3906469888K                 active/Online
         1    21881f31  3906469888K                 active/Online
         2    8296e5e6  3906469888K /dev/sdb        active/Online
         3    8d40b56c  3906469888K                 active/Online

sdc vypadnul jako druhej pote, co zmizel sdb, examine:

      Number    RefNo      Size       Device      Type/State
         0    b854cc92  3906469888K                 active/Online
         1    21881f31  3906469888K /dev/sdc        active/Online
         2    8296e5e6  3906469888K                 active/Offline, Failed, Missing
         3    8d40b56c  3906469888K                 active/Online

sdd vypadnul az po zastaveni pole, jeho examine:

      Number    RefNo      Size       Device      Type/State
         0    b854cc92  3906469888K /dev/sdd        active/Online
         1    21881f31  3906469888K                 active/Offline, Failed
         2    8296e5e6  3906469888K                 active/Offline, Failed, Missing
         3    8d40b56c  3906469888K                 active/Online

totez sde:

      Number    RefNo      Size       Device      Type/State
         0    b854cc92  3906469888K                 active/Online
         1    21881f31  3906469888K                 active/Offline, Failed
         2    8296e5e6  3906469888K                 active/Offline, Failed, Missing
         3    8d40b56c  3906469888K /dev/sde        active/Online

  Stav uzitecnych dat ocekavam stejny, nakolik pole NEbylo odemknuto ani namountovano a vsechny disky jsou citelne (to jsem kontroloval po hlasce o dvou selhanych discich jako prvni, komplet cteni celyho disku pres dd, nez mi doslo, co se stalo).   Takze otazka zni... je bezpecne na tom pustit:

mdadm --assemble --force /dev/md126 /dev/sdd /dev/sdc /dev/sdb /dev/sde

Nebo neco jinyho nejak jinak? Zalezi na poradi disku v tomto prikazu uvedenych, pokud tam vubec maji byt uvedeny?

  Misto na bitovou kopii 16 TB dat nemam. Zalohu aspon tech dulezitych dat pripadne mam, pomene cerstvou a fyzicky offline. Ale samozrejme preferovana varianta je, aby byla slusna sance na to, ze preziji i data mene dulezita.

Diky za napady.