ʻO RAID 5 a i ʻole RAID 6 Hoʻohaʻahaʻa ʻia me ka lua o ka holo ʻana i ka wā o ke kūkulu hou ʻana (No ke aha e holo ʻole ai i ka CHKDSK a i ʻole fsck, ddrescue Sector Cloning, a me mdadm Read-Only Assembly)¶

Ke hoʻololi nei i kahi kīkī paʻa ʻole i loko o kahi pahu i hoʻohaʻahaʻa ʻia RAID 5 a i ʻole RAID 6 kikowaena mālama, Synology DiskStation, QNAP NAS, Linux mdadm array, a i ʻole Dell PERC / HP Smart Array / LSI MegaRAID mea hoʻoponopono lako lako wale nō e nānā i ka ** lua o ka hoʻolei ʻana o nā lālā i koe UNC (Hino Helu Helu ʻole) nā ʻāpana ʻino, Pending Sectors, a i ʻole e waiho i waho ma 42% ʻO ka parity rebuild** kekahi o nā pilikia ulia pōpilikia i loko o ka ʻoihana a me ka computing home-lab. I ka hui hookahi RAID 5 pūʻulu (N - 1 ka hoʻomanawanui hewa), ʻo ke kūkulu hou ʻana o ka makemakika XOR o ka disk pani e pono ai e heluhelu i kēlā me kēia māhele (LBA 0 i ka LBA hope loa) ma waena o 100 pakeneka o ke koena N - 1 holo ka lālā i ka manawa like. I kēia manawa kiʻekiʻe-kaha 8 TB i 24 TB helium drives (kūʻai ʻia mai ka hui hana like ʻana a hoʻopaʻa ʻia i nā hola wela a me nā haʻalulu like) e loaʻa i kahi heluhelu hou-terabyte sequential rebuild heluhelu, hoʻokahi defect media latent (URE) a i ʻole ke poʻo heluhelu nāwaliwali ma kahi kaʻa lua e hoʻopau ai ka mea hoʻokele RAID i ke kūkulu hou ʻana, e kiola i ka lua i waho e like me Faulty / Crashed, a e kaha i ka loko waihona waihona Offline / Crashed / Superblock Event Count Mismatch.

Mai makaʻu, a ** mai hoʻoikaika i ka hana hou ʻana a i ʻole holo i nā pono hana hoʻoponopono filesystem (fsck, e2fsck, xfs_repair, a i ʻole Windows CHKDSK)**. ʻOiai ke kaʻa lua i loko RAID 5 (a i ʻole ka holo ʻekolu i loko RAID 6) ulu 50 i 5,000 ʻino a kipaku ʻia i waho o ka papa, **99.9999 pākēneka o nā ʻāpana ma kēlā "hāʻule" lua ka holo ʻana he 100% hiki ke heluhelu ʻia **. Ma ka hooki koke ʻana i ka pūʻulu, hoʻopaʻa ʻana i kēlā me kēia lālā paʻa ʻole i ka ʻāpana disk i kēlā me kēia ʻāpana i nā kaʻa hoʻomaha olakino e hoʻohana ana i GNU ddrescue, a me ka hui ʻana i nā kiʻi cloned i Read-Only / --assume-clean ke ʻano uhi (mdadm + dmsetup Kopi-ma-Kākau i nā uhi uhi), hiki iā ʻoe ke kau i ka waihona me ka palekana a kope i nā faila āpau. No nā kikoʻī kikoʻī o nā mea hana NAS a me nā hemahema o ka mea hoʻokele waiwai, e heluhelu i kā mākou alakaʻi alakaʻi ʻO RAID 5 a i ʻole RAID 6 Synology, QNAP, a i ʻole TrueNAS hāʻule kaʻa he nui, ʻO ka make kāleka hoʻoponopono RAID a me ka hoʻokomo ʻana i nā ʻāina ʻē, a Ua hoʻohaʻahaʻa ʻia ka pūʻulu NAS RAID a me ka hoʻihoʻi ʻana o ZFS/Btrfs. Inā pili ka hemahema o kāu kikowaena i nā ʻōnaehana palekana ola a i ʻole nā ​​mea hana olakino, e kelepona i kāu helu ulia pōpilikia kūloko (911 ma ʻAmelika Hui Pū ʻIa a me Kanada, 112 ma waena o ka European Union, 999 ma United Kingdom) a i ʻole e nānā i kā mākou papa kuhikuhi o helu pilikia.

E hana i kēia i kēia manawa¶

1. Hoʻopau i ka hana hou ʻana a i ʻole ka holoi ʻana i ka hui a hoʻopau koke i ka laina (a i ʻole e kāpae i ka mana SATA/SAS backplane inā e hana ana kahi kaʻa i ke kaomi ʻana a i ʻole ke kani ʻana i nā leo):
2. No ke aha e hoʻomau ai ka hoʻomau ʻana i ke kūkulu hou ʻana e hoʻopau i ka laha: Ke pā ka mea hoʻokele lālā i kahi ʻāpana ʻino (LBA pau ka manawa heluhelu) a i ʻole he poʻo heluhelu/kākau haʻahaʻa (media scratch), ke waiho nei i ka mea hoʻoponopono RAID e hoʻāʻo hou ana i nā ʻāpana no nā hola e wili i ka microscopic magnetic platter i uhi ʻia i ka lepo ma o ka pahu holoʻokoʻa holoʻokoʻa, e hoʻohuli i kahi mea hiki ke hoʻihoʻi. 20-sector kīnā i loko o kahi pōʻino poʻo poʻo pā piha.
3. **AOLE holo CHKDSK /f, fsck -y, e2fsck, btrfs check --repair, a i ʻole xfs_repair ma kahi ʻano RAID i hoʻohaʻahaʻa ʻia, ʻaʻohe o ka hana, a i ʻole ka hapalua i kūkulu hou ʻia **:
4. No ke aha mai CHKDSK a fsck hoʻopau i ka 100% ʻikepili mau loa ma waho o-sync RAID: Inā hāʻule ʻo Drive 2 i ka Pōʻalua a hāʻule ʻo Drive 3 i ka hana hou ʻana o ka Pōʻalima, mdadm a i ʻole RAID lako Event Counter (Update Time) ma ia mau kaʻa ʻelua ʻokoʻa e nā kaukani o nā kālepa. Inā ʻoe e hōʻuluʻulu ikaika i kahi pūʻulu me nā hanauna hoʻokele like ʻole a i ʻole nā ​​paʻi nalo a ʻae fsck a i ʻole CHKDSK no ke kākau ʻana i ka laha, ʻike ka mea nānā filesystem i nā kumu lāʻau inode i haki ʻia e nā kaha RAID i kuhi hewa ʻole, ** holoi i nā miliona o nā palapala kuhikuhi kūpono i loko lost+found, a kākau hou i nā kaha ʻikepili maikaʻi me ka parity ʻino ma kēlā me kēia disk i loko o ka array**.
5. AOLE kaomi "Create Array", "Initialize", a i ʻole "Clear Configuration" ma kahi lako RAID BIOS, a ʻAʻole holo mdadm --create me ka ole --assume-clean a me kahi kope kope (COW) uhi uhi:
6. Holo i kahi maʻamau mdadm --create (me ka ʻole --assume-clean a me ka hoʻohālikelike ʻana i ka nui o ka puʻupuʻu kikoʻī, ka hoʻonohonoho ʻana, a me ka offset data) e hoʻomaka koke i ke kākau ʻana i nā paʻi parity hou ma nā disks āpau, e luku ana i ka filesystem superblock a me B-lāʻau inā loaʻa kekahi ʻāpana (512K vs. 64K ʻāpana, left-symmetric vs. left-asymmetric, a i ʻole 2048s vs. 262144s ʻikepili offset) hewa ʻole e hoʻokahi ʻāpana.
7. E hoʻopaʻa inoa i kēlā me kēia ʻīpuka paʻa paʻa kino (Bay 0 / Bay 1 / Bay 2 / Bay 3 a me ka helu serial) me kahi hōʻailona mau a i ʻole lipine MAMUA o ka huki ʻana i hoʻokahi kaʻa i waho o ka chassis:
8. E kākau i ka helu bay pololei a me ka helu serial ma ka lepili luna o kēlā me kēia kaʻa. ʻIke i ke kauoha bay kino a me ka nānā ʻana mdadm --examine Mālama ka metadata i nā hola o ke kūkulu hou ʻana i ke kauoha ma hope.
9. ** Hoʻopaʻa i ka (mau) kaʻa holo hewa ʻole i kēlā me kēia ʻāpana i nā drive hou e hoʻohana ana GNU ddrescue (me kahi palapala palapala) ma luna o nā awa SATA/HBA motherboard ma mua o ka hoʻāʻo ʻana i kekahi hui RAID**:
10. Mai ho'āʻo e hōʻuluʻulu a kau i kahi pūʻulu RAID me ka hoʻohana ʻana i ke kaʻa make kino. E hoʻohui i ka holo hemahema i kahi non-RAID SATA/AHCI a i ʻole LSI IT-Mode ʻO ke awa HBA ma kahi hana hoʻopakele Linux, hoʻopau i ka hoʻouka ʻana o OS, a holo ddrescue -f -n /dev/sdX /dev/sdY rescue_driveX.map no laila ua hoʻokuʻu ʻia nā ʻāpana wikiwiki olakino ma mua o ka pā ʻana i nā ʻāpana ʻino.
11. ** E hōʻuluʻulu i nā pahu cloned i loko Read-Only ʻano (mdadm --assemble --readonly a hoʻohana paha dmsetup / losetup RAM/ waihona i uhi ʻia)**:
12. Ma ke kau ʻana i kahi kope kope-ma-Kākau (COW) e kau i ka faila ma luna o kēlā me kēia diski lālā cloned, hiki iā ʻoe ke hoʻāʻo-mauna i ka array, hoʻokani hou i kahi ext4 a i ʻole XFS puke pai, a i ʻole e hoʻāʻo i ke kūkulu hou ʻana ** ʻoiai e hōʻoiaʻiʻo ana ʻaʻole i hoʻololi ʻia nā byte ʻole ma kāu mau ʻāpana diski lalo**.

ʻO ka lāʻau hoʻoholo: hoʻopakele i kahi pūʻulu RAID 5 a i ʻole RAID 6 i ka wā i hāʻule ai ka lua o ke kaʻa i ke kūkulu hou ʻana.¶

E hahai i kēia kaʻina hana hoʻolaʻa mālama forensic e hoʻihoʻi hou i kāu pūʻulu me ka nalowale ʻole o hoʻokahi kaula ola:

flowchart TD
    Start["ʻO RAID 5 (2nd Drive) a i ʻole RAID 6 (3rd Drive) hāʻule / hāʻule pahemo i ka wā o ke kūkulu hou ʻana."] --> StopNow["KANAWAI 1: Hoʻopau koke i ke kūkulu hou ʻana a me ka hoʻohaʻahaʻa ʻana i ka lālani (E pale i ka helu ʻana i ka pā ma ka holo ʻole ʻana)"]

    StopNow --> SoundCheck{"Ua hana ʻia ke kaomi nui ʻana, ka bip, a i ʻole ka wili ʻana i nā kani?"}

    SoundCheck -->|"'Ae: Po'o Mechanical Po'o Po'o / Pa'i"| CleanRoom["STOP: Mai hana hou i ka holo. E hoʻouna WALE i ke Kaʻa Mechanically Failed (Plus 1 Healthy Drive for Stripe Reference) i kahi lumi hoʻomaʻemaʻe e kiʻi ai i kahi clone"]

    SoundCheck -->|"ʻAʻole: Holo mālie, hāʻule ma muli o nā ʻāpana maikaʻi ʻole (UNC / Ke kali ʻana / Timeout)"| LabelDrives["NĀ HANA 2: E lepili i kēlā me kēia kaʻa me kāna helu Bay pololei a me ka helu Serial -> Hoʻohui i ka Linux Workstation ma o Direct SATA / IT-Mode HBA"]

    LabelDrives --> InspectMeta["Holo 'mdadm --examine /dev/sd*' a me 'smartctl -a /dev/sd*' ma nā Drives a pau (E hoʻopaʻa i ka manawa hou, helu helu hanana, kuleana, ka nui o ka Chunk, Layout, Data Offset)"]

    InspectMeta --> CloneDrives["KANAWAI 3: Hoʻopili i kēlā me kēia kaʻa hoʻohaʻahaʻa / ʻino-sector i kahi kaʻa hoʻomaha maikaʻi me ka hoʻohana 'ddrescue -f -n' (Fast Pass Mua, A laila Scrape Pass me Mapfile)"]

    CloneDrives --> PickFreshest["E hoʻomaopopo i ka N-1 (RAID 5) a i ʻole N-2 (RAID 6) me ka helu 'Hana' HOʻOPIʻI KIʻekiʻe / LOA loa (Mai hoʻohui i kahi Kaʻa Kahiko i hāʻule ʻia me nā pahu hou)"]

    PickFreshest --> CowOverlay["KaʻAnuʻu Hana 4: E hana i ka heluhelu-wale Loop / 'dmsetup' Copy-on-Write (COW) overlays ma luna o nā Cloned Drives No laila ʻaʻole hiki ke hoʻololi ʻia nā Clones kumu."]

    CowOverlay --> AssembleRO["E holo i ka 'mdadm --assemble --force --readonly' ma luna o nā polokalamu Overlay -> Mount Filesystem Read-Only ('mount -o ro,noload') -> E kope i nā ʻikepili koʻikoʻi āpau i ka waihona waho."]

Māhele 1: Ka hoʻomaopopo ʻana i ke kumu i hāʻule ʻole ai ka lua i ka wā o kahi kūkulu hou ʻana o RAID (XOR Parity Math & Event Counters)¶

No ka hoʻomaopopo ʻana i ke kumu o ka hāʻule ʻana o kāu pūʻulu ma waena o ke kūkulu hou ʻana a no ke aha e hoʻihoʻi hou ʻia ai kāu ʻikepili, e nānā pehea RAID 5 a RAID 6 hana maoli ma ka diski.

1. Pehea XOR hana hou ʻia parity a no ke aha 1 hoʻopau ka ʻāpana kino i ke kūkulu hou ʻana¶

I ka 4-kaa RAID 5 pūʻulu (Drive A, Drive B, Drive C, Drive D), hoʻokaʻawale ʻia ka ʻikepili i nā chunks nui paʻa (maʻamau 64 KB a i ʻole 512 KB) i kāʻei ʻia ma nā diski, kahi e paʻa ai ka XOR Parity (P): $\(\text{Parity } P = A \oplus B \oplus C\)$ Ina Drive C hāʻule loa a hoʻokomo ʻoe i kahi hakahaka Replacement Drive C_new, pono e hana hou ka mea hoʻoponopono RAID i kēlā me kēia kaha C_new ma ka helu ʻana: $\(C_{\text{kūkulu hou}} = A \oplus B \oplus P\)$ - ** He aha ka mea e hana ai i ka wā Drive B paʻa i kahi hiki ʻole ke heluhelu ʻia 4 KB ʻāpana (UNC hewa) ma 42% o ke kūkulu hou ʻana**: - No ka helu ʻana i kēlā pākahi 4 KB poloka ma C_new, ka mea hoʻoponopono * pono * loaʻa kahi heluhelu kūpono mai Drive A, Drive B, a Drive D. - I ka manawa Drive B hoʻihoʻi i kahi lako Read Error (Sense Key 03h / 11h Unrecovered Read Error) ma hope o 7 kekona (TLER / ERC hoʻopau manawa), hoʻolālā pinepine ʻia nā mea hoʻokele RAID o ka mea kūʻai aku a me ka ʻoihana me kahi ʻano paʻamau pōʻino: ** ma kahi o ka lele ʻana i kēlā 4 KB ʻāpana, hoʻopaʻa inoa i kahi poloka zero, a hoʻopau i kekahi 99.9999% o ke kūkulu hou ʻana, hōʻailona ka mana RAID Drive B e like me FAILED / OFFLINE a hoʻopau i ke kūkulu hou ʻana **. - I kēlā manawa, loaʻa i kāu hui he ʻelua pahu i hōʻailona ʻia "Holo" (Drive C ʻo ia wale nō 42% kukulu hou ia, a Drive B i loaʻa maoli 99.99998% nā ʻāpana olakino a wale nō 12 nā ʻāpana ʻino).

2. No ke aha Event Counters (Update Time) e hoʻoholo i nā pahu e pono ai ʻoe e hoʻohana no ka hoʻihoʻi¶

I kēlā me kēia manawa a RAID 5 a i ʻole RAID 6 kākau ʻo array i ka ʻikepili a hoʻololi paha i ke kūlana (e like me ka hāʻule ʻana o kahi kaʻa), ka metadata RAID (mdadm superblock a mea lako paha DDF metadata) hoʻonui i kahi 64-bit Events counter ma nā hoʻokele hana a pau i kēia manawa.

Hoʻohālikelike i ka mdadm --examine puka o ka 4-drive maʻamau RAID 5 pōʻino ma hea sdb hāʻule i nehinei (Events: 10450) a sdc haʻalele i ka wā o ke kūkulu hou ʻana i kēia lā (Events: 10892):

• Ke Kumu gula o RAID Recovery mai ka papa ma luna:
• E hoʻomaopopo i kēlā /dev/sda (10892), /dev/sdc (10890), a /dev/sdd (10892) ua like like nā mea a pau Events helu helu mai Today 14:22.
• I ka manawa mākou clone /dev/sdc ma luna o kahi kaʻa olakino e hoʻohana ana ddrescue (e hoʻihoʻi ana i ka 99.9999% o kāna mau ʻāpana a me ka hoʻopiha piha ʻana i nā ʻāpana 8 hiki ʻole ke heluhelu ʻia), hiki iā mākou ke hoʻohui ikaika /dev/sda + Cloned /dev/sdc + /dev/sdd (ka nānā ʻole i ka hapalua i kūkulu hou ʻia /dev/sdb loa). Hāʻawi kēlā hui 3-drive iā mākou i ka 100 pākēneka kūlike, hoʻohaʻahaʻa ʻia a hiki i ka lua RAID 5 pūʻali me ka haehae ʻole ʻana.

Māhele 2: ʻO ka hoʻopili ʻana i ka ʻāpana i kēlā me kēia ʻanuʻu o ka holo hemahema me GNU ddrescue¶

Ma mua o kou hoʻopā ʻana mdadm a i ʻole kekahi mea hana RAID hui, pono ** ʻoe e hoʻopaʻa i ka drive (mau) i hoʻolei i nā hewa heluhelu i nā pahu hoʻololi olakino o ka nui a i ʻole ka nui byte.

KaʻAnuʻu Hana 1: E hoʻomākaukau i ke kaiapuni hoʻopakele Linux a hoʻopau i ka auto-assembly¶

1. E hoʻopaʻa i kahi keʻena hana Linux (a i ʻole e hoʻopaʻa i kāu kikowaena mai kahi lāʻau USB ola e like me SystemRescue system-rescue.org a i ʻole Ubuntu Live Server) me nā drive e pili ana i ka maʻamau ** motherboard SATA ports ** a i ʻole LSI HBA ma IT Mode (ʻAʻole-RAID pass-through).
2. E pale iā Linux mai ka hoʻāʻo ʻana e hōʻuluʻulu a kau i kāu pūʻali RAID i haki ʻia ma ka boot:
3. Ke hoʻomaka nei ʻo SystemRescue a i ʻole Ubuntu Live, e nānā cat /proc/mdstat. Inā hui pū ʻo Linux i kahi ʻāpana md mea ma ke kua, e oki koke ia ma ke ano palekana heluhelu-wale ma mua o ke kii ana:

   mdadm --stop /dev/md*
   

4. E ʻike i kēlā me kēia kaʻa ma ke kumu hoʻohālike pololei, ka helu serial, a me ka nui byte (mai kuhi /dev/sdX nā leka, no ka mea, hāʻawi hou ʻo Linux i nā leka hoʻokele ma waena o nā reboots):

   lsblk -o NAME,SIZE,MODEL,SERIAL,TYPE,FSTYPE
   

   E kakau iho i ka mea /dev/sdX e pili ana i kāu Failing Source Drive (no ka laʻana, /dev/sdc, serial WCC4E12345) a ʻo wai /dev/sdY ʻo kāu Brand-New Blank Target Drive (no ka laʻana, /dev/sdf, serial ZGG9988776).

KaʻAnuʻu Hana 2: Holo i ka 2-Stage GNU ddrescue cloning protocol (E hoʻohana mau i kahi palapala palapala)¶

Mai hoʻohana maʻamau dd (dd if=/dev/sdc of=/dev/sdf) ma kahi kaʻa RAID hemahema: maʻamau dd a hoʻopau paha i ka ʻāpana ʻino mua a i ʻole e hoʻololi i nā offset pukana inā sync ua kuhi hewa ʻia, a ʻaʻohe ona hoʻomanaʻo i nā ʻāpana i kūleʻa. GNU ddrescue (gddrescue package) hoʻohana i kahi palapala palapala mokuʻāina multi-pass naʻauao:

1. Pass 1: ʻO Clone ʻAʻole Hoʻopaʻa ʻole (-n hae) e hoʻopakele iā 99.99% o nā ʻāpana olakino mua:

   ddrescue -f -n -b 4096 /dev/sdc /dev/sdf /root/raid_bay3_sdc.map
   

2. -f: ʻAe ʻia ke kākau pololei ʻana i ka mea hana /dev/sdf (ʻike ʻekolu i kēlā /dev/sdc ʻo ia ke kumu a /dev/sdf ʻo ia ka pahu hopu hakahaka).
3. -n (--no-scrape): Haʻi ddrescue e lele koke ma luna o kekahi māhele lohi a maikaʻi ʻole me ka wili ʻole i ke poʻo heluhelu, e hoʻopakele i nā gigabytes olakino āpau i ka wikiwiki loa (150 to 250 MB/s) ma mua o ka piʻi ʻana o ke olakino o ke kaʻa.
4. /root/raid_bay3_sdc.map: Mālama i ka holomua ʻāpana pololei i kēlā me kēia 30 kekona. ʻOiai inā paʻa ka pahu hemahema a koi i ka pōʻaiapuni mana, e holo hou i ke kauoha like me ka .map Hoʻomaka hou ka faila ma ka ʻāpana pololei kahi i kū ai.
5. Pass 2: Hoʻopaʻa ʻia i ka hoʻihoʻi ʻana/ʻokiʻoki ʻia (-d -r1) e hoʻihoʻi i nā ʻāpana ʻino ʻole:
6. Ke pau ka hala 1 (pct rescued: 99.99%), e hoʻohaʻahaʻa i ka kernel SCSI kauoha manawa no ka pau ʻole o kēlā me kēia ʻāpana ʻino (5 seconds ma kahi o ka kali ana 30 to 180 seconds no kēlā me kēia ʻāpana):

   echo 5 > /sys/block/sdc/device/timeout
   

7. Holo i ka Pass 2 me ka loaʻa ʻana o ka disk (-d) a 1 ho'āʻo hou (-r 1):

   ddrescue -f -d -r 1 -b 4096 /dev/sdc /dev/sdf /root/raid_bay3_sdc.map
   

8. ** E pani i lalo, e wehe i ka mea mua i hāʻule /dev/sdc kaʻa, a hoʻonoho maluhia ma kahi papa**: Mai kēia manawa aku, e hana wale ʻoe me kāu clone olakino (/dev/sdf) a me kāu mau lālā olakino ʻē aʻe.

Māhele 3: Heluhelu-wale i ka RAID hui me ka hoʻohana ʻana i Copy-on-Write (COW) uhi a mdadm¶

I kēia manawa ua loaʻa iā ʻoe nā clones disk olakino, ʻaʻole holo mdadm --assemble --force pololei ma luna o nā kaʻa maka me ka pale ʻole ʻana i ka Copy-on-Write (COW) uhi a i ʻole --readonly hae. No ke aha mai? No ka mea --force kākau hou i nā superblocks RAID ma nā kaʻa, a inā he waihona waihona waihona (ext4 / XFS) hoʻokani hou ma ka mauna me ka hoʻohana ʻana i ka hui hoʻokele hewa, hoʻololi mau ia i kāu mau clones.

KaʻAnuʻu Hana 1: E nānā i ka RAID metadata ma waena o nā lālā a pau (mdadm --examine)¶

Holo mdadm --examine ma kēlā me kēia lālā hoʻokele (a i ʻole ʻāpana, e laʻa, /dev/sda1 a i ʻole /dev/sda3 ma Synology/QNAP NAS drives) a mālama i ka hopena i kahi faila kikokikona:

mdadm --examine /dev/sd[a,b,d,f]* > /root/raid_metadata_audit.txt
cat /root/raid_metadata_audit.txt

KaʻAnuʻu Hana 2: Hana i mea hoʻopōʻino ʻole i ke kope-ma-Kākau (COW) Nā mea hoʻopololei-Mapper (overlay-create)¶

E hōʻoiaʻiʻo i ka palekana 100% no laila inā ʻoe e hoʻohui pū i ka hui a holo mount, ** ʻaʻohe paita i kākau ʻia i kāu mau paʻa paʻa kino**, hana i ka loop-backed dm-snapshot Kope-ma-Kākau i nā uhi i loko /tmp (a i ʻole ma kahi kaʻa kaʻa waho waho):

# Create a small sparse COW file + loop device for each member partition (e.g., sda3, sdf3, sdd3)
mkdir -p /root/cow
for dev in sda3 sdf3 sdd3; do
    truncate -s 20G /root/cow/${dev}.cow
    loop_base=$(losetup --find --show --read-only /dev/${dev})
    loop_cow=$(losetup --find --show /root/cow/${dev}.cow)
    sz=$(blockdev --getsz /dev/${dev})
    echo "0 $sz snapshot $loop_base $loop_cow P 8" | dmsetup create ovl_${dev}
done

KaʻAnuʻu Hana 3: E hōʻuluʻulu ikaika i ka papa ma luna o nā mea i uhi ʻia a kau i ka Helu-wale¶

1. ** E hōʻuluʻulu i ka mea hou loa N - 1 e hoohana ana i na mea hana uhi --assemble --force**:

   mdadm --assemble --force --verbose /dev/md99 /dev/mapper/ovl_sda3 /dev/mapper/ovl_sdf3 /dev/mapper/ovl_sdd3
   

2. No ke aha mākou e haʻalele ai i ke kaʻa paʻa a kūkulu hou ʻia (sdb3): Ke koi ʻana mdadm e hui pū me nā 3 mea hou i kēia lā (sda3, hoʻopaʻa ʻia sdf3, a sdd3) lawe mai /dev/md99 pūnaewele ma ke ʻano hoʻohaʻahaʻa maʻemaʻe ([U_UU]) me ka hoʻohana ʻana i ka mea hou loa Events mokuʻāina, ʻoiai ddrescue ua hoopakele i ka ikepili mai sdf3.
3. E ho'ā i ka LVM (vgchange) inā hoʻohana kāu NAS a i ʻole kikowaena iā LVM ma luna o RAID (Synology SHR, QNAP, Proxmox, Ubuntu):

   pvscan
   vgscan
   vgchange -ay
   lvs
   

   (Inā hoʻohana kāu Synology NAS Btrfs ma LVM a i ʻole Synology Hybrid RAID, e ike auanei oe /dev/vg1000/lv a i ʻole /dev/vg1/volume_1 puka mai.)
4. E kau i ka waihona ma ke ʻano Heluhelu-wale wale nō a e haʻalele koke i kāu mau faila koʻikoʻi:

   mkdir -p /mnt/raid_rescue
   # For ext4:
   mount -o ro,noload /dev/md99 /mnt/raid_rescue
   # (Or if noload refuses due to dirty journal, mount the COW overlay normally so the journal replays into the disposable COW file)
   mount -o ro /dev/md99 /mnt/raid_rescue
   # For Btrfs (Synology NAS):
   mount -t btrfs -o ro,usebackuproot,nologreplay /dev/vg1000/lv /mnt/raid_rescue
   

5. Kopi mua i kāu mau waihona koʻikoʻi rsync i kahi kaʻa hoʻihoʻi waho:

   rsync -avP --ignore-errors /mnt/raid_rescue/ /mnt/external_backup_usb/
   

Māhele 4: Hoʻopakele ʻana i nā mea hoʻokele RAID Hardware (Dell PERC, LSI MegaRAID, HP Smart Array, Adaptec) a me nā Pool ZFS / Btrfs¶

He aha inā inā ʻaʻole i mālama ʻia kāu array e Linux mdadm, akā na kahi Kāleka Hoʻoponopono RAID Paahana (Dell PERC H730/H740/H755, Broadcom / LSI MegaRAID, a i ʻole HP Smart Array) e hōʻike nei Virtual Disk Offline aiʻole Foreign Configuration Found ma hope o ka lua o ka hoʻokele ʻana i ka wā e kūkulu hou ai?

1. Nā Kāleka RAID Lako Paʻa (Dell PERC, LSI MegaRAID, HP Smart Array)¶

1. AOLE kaomi "Clear Foreign Configuration" a i ʻole "Initialize Virtual Disk":
2. Ma Dell PERC / LSI MegaRAID BIOS (Ctrl+R a i ʻole HII UEFI Configuration Utility), Clear holoi i ka metadata RAID mai nā drive, a Initialize kākau hou i nā kaha ʻikepili.
3. Inā hoʻokuʻu ʻia kahi kaʻa olakino ma waho ma muli o kahi uwea holo hope a i ʻole ka manawa pau, Import Foreign Configuration ho'āʻo e hoʻihoʻi i ke kaʻa i loko; ** akā naʻe, inā hāʻule kekahi kaʻa i nā pule i hala aku nei a hāʻule ka lua i kēia lā, hiki i ka lawe ʻana i nā mea ʻelua a i ʻole ke koi ʻana i ka drive stale ma ka pūnaewele ma o ka mana BIOS hiki ke hōʻino i ka array**.
4. Ke ʻano hana hoʻoponopono polokalamu palekana no Hardware RAID (DDF Metadata / dmraid / R-Studio / UFS Explorer):
5. Paʻa hale kūʻai kāleka RAID paʻa SNIA DDF (Disk Data Format) metadata ma nā ʻāpana hope loa ma ka hopena o kēlā me kēia pahu paʻakikī, a e hoʻokaʻawale i nā ʻikepili ma waena o nā pahu e hoʻomaka ana ma ka ʻāpana 0 (a i ʻole ma hope o kahi poʻo liʻiliʻi offset) me ka hoʻohana maʻamau 64 KB, 128 KB, 256 KB, a i ʻole 512 KB ka nui o nā ʻāpana (Left-Asymmetric a i ʻole Left-Symmetric).
6. Ma kahi o ka pāʻani me kahi mea hoʻopōʻino RAID kāleka BIOS:
   1. E lepili i nā drive āpau ma ka helu slot a hoʻohui iā lākou i kahi ** SATA / non-RAID maʻamau IT-Mode HBA** ma ke kahua hana ho'ōla.
   2. Hoʻopili i ka hoʻokele ʻino me ka GNU ddrescue e like me ka mea i hōʻike ʻia ma ka Phase 2.
   3. E wehe i nā kaila cloned ma UFS Explorer RAID Recovery aiʻole R-Studio Technician, a i ʻole e hōʻuluʻulu iā lākou i heluhelu-wale ma Linux me ka hoʻohana ʻana dmsetup / mdadm --build: ʻike pono kēia mau mea hana i ka LSI / PERC / HP kaʻina kaha, ka nui o ka ʻāpana (64K/512K), a me ka hoʻololi ʻana i ka parity i 100 pakeneka heluhelu-wale nō me ka ʻole e ʻae i ke kāleka RAID hāmeʻa e kākau i hoʻokahi byte i kāu mau drive.

2. TrueNAS / ZFS (RAID-Z1 / RAID-Z2) a me ka Synology Btrfs Holoi ʻia ka ʻōpala/Resilver¶

Inā hoʻohana kāu loko i ka OpenZFS (RAID-Z1 a i ʻole RAID-Z2) ma TrueNAS a i ʻole Proxmox a me ka lua o ka pahu pahu CKSUM / READ hewa i ka wā a zpool replace kalahala: 1. **No ke aha ʻo ZFS e mālama ai i nā ʻāpana ʻino i ka wā e hoʻihoʻi hou ai i ʻoi aku ka maikaʻi ma mua o ka RAID lako, a i ka wā e laka mau ai : - ʻAʻole like me nā kāleka RAID hāmeʻa naive e hāʻule i kahi disk holoʻokoʻa ma kahi ʻāpana maikaʻi ʻole hoʻokahi, hoʻomau ka ZFS i ka hoʻihoʻi hou ʻana ma mua o ka localized. READ hewa a hoʻopaʻa wale i nā ala faila i hōʻino ʻia ma zpool status -v. - Eia nō naʻe, inā e kau ka lua o ka holo hewa i ka pahi SATA/SAS (I/O timeout / poʻo hahau) a hāʻule paha i waho (UNAVAIL / FAULTED), komo ka loko SUSPENDED mokuʻāina (pool I/O is currently suspended). 2. Pehea e hoopakele ai a SUSPENDED a i ʻole FAULTED ZFS RAID-Z1 / RAID-Z2 loko: - E hoʻohaʻahaʻa me ka maʻemaʻe, e hoʻopili i ka drive hemahema ma kahi disk hoʻololi maikaʻi me GNU ddrescue -f -n (ʻO nā lepili ʻāpana ZFS me ka pololei vdev GUID ma ka hoʻomaka a me ka hopena o ka disk, no laila, ʻike ʻia kahi clone ʻāpana maka e ZFS ma ke ʻano he kumu. vdev lālā). - Hoʻokaʻawale i ke kaʻa make kino, haʻalele i ka ddrescue Hoʻopili ʻia ka clone, a hoʻokomo i ka wai ZFS ma ** ke ʻano Heluhelu wale nō (readonly=on) me ka hoʻomaka ʻana i kahi kālepa kākau a hoʻomau i ka resilver:

zpool import -o readonly=on -f -R /mnt/zfs_rescue tank

Nīnau pinepine¶

No ke aha ʻoe i ʻōlelo ai ʻAʻole e holo CHKDSK a i ʻole fsck i ka wā e piʻi ai kaʻu pūʻulu RAID me nā hewa filesystem ma hope o ka hāʻule ʻana o ka drive?¶

Ma kahi paʻakikī kūʻokoʻa hoʻokahi, e holo ana CHKDSK a i ʻole fsck hoʻoponopono i nā ʻano like ʻole o ka papa kuhikuhi liʻiliʻi. Ma kahi multi-drive RAID 5 a i ʻole RAID 6 pūʻulu i loaʻa i ka hāʻule pālua a i ʻole ke kūkulu hou ʻana, ʻaneʻane i nā manawa a pau nā hewa ʻōnaehana waihona ʻO RAID stripe misalignment a i ʻole kahi kaʻa lālā paʻa i koi ʻia ma ka pūnaewele (ʻo ia hoʻi kēlā me kēia 3 a i ʻole 4 64 KB Heluhelu ʻia ka ʻāpana o ka ʻikepili mai ka hana hoʻokele hewa). Inā ʻoe e ʻae CHKDSK /f a i ʻole fsck -y E holo ana ma kahi laina RAID i kuhi hewa ʻole a i ʻole ka hui pū ʻana, kuhi hewa ʻo ia i nā miliona o nā faila kūpono e like me ka metadata hewa, ʻoki i nā lāʻau papa kuhikuhi, hoʻolei i nā ʻāpana inoa ʻole i loko. nalowale+loaʻa, a kākau i nā poloka parity hou ma nā kaʻa a pau, no laila ʻaʻole hiki ke hoʻohui hou i ka hui hoʻokele pololei ma hope. E kau mau ma o Heluhelu-wale ke kope-ma-kakau (BIPI) nā uhi mua.

He aha inā inā i kūkulu hou ʻia ka pahu pani (42% piha) A me ka lua o ka holo i hāʻule ma 42% Ua pono nā ʻikepili ʻelua? Hiki iaʻu ke hoʻohui iā lākou?¶

ʻAe. ʻO kēia kahi hoʻopakele "Split-Horizon / Hybrid Clone" RAID. Ina paha Ke kalaiwa 2 make loa i ka Poakahi, hookomo oe Kaʻa Hou 2, a ua hana hou ia 42% (LBA 3,500,000,000) mamua Ke kalaiwa 3 ua loaʻa i kahi pōʻino poʻo make ma LBA 3,500,000,000. E nānā i kāu mea maoli: no ka 42% mua o ka papa (0% i 42%), Kaʻa Hou 2 ua hana hou ʻia he 100 pakeneka. No laila no ka mea mua 42% o ka laulā LBA, loaʻa iā ʻoe kahi hoʻonohonoho piha o nā kaha Ka‘a 1 + Ka‘a Hou 2 + Ka‘a 4 (hāʻule ka mālama ʻana Ke kalaiwa 3 e like me ka diski nalowale). A ina Ke kalaiwa 3 Ua loaʻa wale i kahi ʻāpana ʻino i loko 42% (a hiki ke heluhelu ʻia mai 43% i 100% ma o ddrescue), hoʻohana ʻoe Ke kaʻa 1 + Kaʻa 3 + Kaʻa 4 no ke koena 58% o ka papa.

No ke aha e hāʻule pinepine ai nā kaʻa lua ma ka waena o kahi kūkulu hou ʻana o RAID 5?¶

Hoʻohui ʻia ʻelua mau mea ʻenehana e hana RAID 5 kūkulu hou i kaulana no ka hāʻule ʻana o ka pahu pālua: (1) ʻO ka ʻelemakule pūʻulu like: Kūʻai ka hapa nui o nā mea hoʻohana i 4 a 8 mau pahu paʻakikī like ʻole mai ka hale kūʻai like ma ka lā hoʻokahi (ʻO ka firmware hoʻokahi, ka nui o ka hale hana like, ka pūʻulu kaʻa spindle like) a holo ia mau ʻaoʻao ʻaoʻao ma ka mahana like pono no 35,000 hola; i ka pau ana o ka Drive 1, he mau la kaawale o Drive 2 a me Drive 3 mai ka paepae komo mechanical hookahi. (2) Nā hewa heluhelu hiki ʻole ke hoʻoponopono ʻia (URE): Ma lalo o ka hana maʻamau i kēlā me kēia lā, 95% ʻAʻole heluhelu ʻia nā ʻāpana waihona anu o kahi pahu paʻa no nā mahina i ka manawa, no laila ʻaʻole i ʻike ʻia nā ʻāpana ʻino huna. A RAID 5 hana hou i ka ikaika a mau 100% Heluhelu ma luna o kēlā me kēia māhele ma kēlā me kēia huakaʻi ola (40 a 100 terabytes o ka heluhelu mau ʻana), e hōʻike koke ana i nā ʻāpana ʻino huna i hōʻiliʻili ʻia i ka makahiki i hala.

Pehea wau e ʻike ai i nā faila i hoʻopilikia ʻia e nā ʻāpana maikaʻi ʻole hiki ke heluhelu ʻia ddrescue pono e hoʻopiha ʻole?¶

GNU ddrescue loaʻa i kahi pono forensic i kūkulu ʻia i kapa ʻia ddrescuelog a ddru_ntfsfindbad / debugfs e palapala ʻāina i ka ʻāpana LBA i ʻike ʻole ʻia mai kāu .palapala ʻāina waihona (kūlana = '-' nā ʻāpana ʻino) ma o ka makemakika lālani RAID pololei i ka papaʻaina inode filesystem. Ke ike oe i ka mea 4 KB ʻAʻole i hoʻopiha ʻia nā poloka i ka wā cloning, hiki iā ʻoe ke nīnau debugfs (icheck a ncheck ma ext4) a i ʻole btrfs inspect-internal logical-resolve e paʻi i ka papa inoa pololei o 2 a i ʻole 3 filenames i uhi ʻia i kēlā mau ʻāpana maikaʻi ʻole, e hōʻoia ana me ka 100 pākēneka makemakika ʻoiaʻiʻo ʻo kēlā me kēia faila ma kāu kikowaena multi-terabyte ʻaʻole i palaʻino.

Nā alakaʻi pili¶

• ʻO RAID 5 a i ʻole RAID 6 Synology, QNAP, a i ʻole TrueNAS i hāʻule i nā drive he nui a no ke aha ke koi ʻana i kahi kūkulu hou e luku ai i ka ʻikepili.
• ʻO ka make kāleka hoʻokele RAID, ka hoʻokomo ʻana i ka hoʻonohonoho haole, a no ke aha ʻaʻole pono ʻoe e hoʻomaka i nā disks
• Ua hoʻohaʻahaʻa ʻia ka pūʻulu NAS RAID, ka hemahema o nā kaʻa he nui, a me ka hoʻihoʻi ZFS/Btrfs
• ʻAʻole paʻa ka paʻakikī a me ka hoʻihoʻi ʻana i ka ʻikepili
• Kākoʻo kamepiula: 3-2-1 hoʻolālā a me ka hoʻōla pōʻino
• Nā helu pilikia ma ka ʻāina

Nā kumu a me ke alakaʻi kūhelu¶

• Linux Raid Wiki (raid.wiki.kernel.org/index.php/Linux_Raid & Recovering_a_failed_software_RAID): Kele Kūlana mdadm Hoʻoponopono Hoʻolaʻa, Mea-Mapper Copy-on-Write (dm-snapshot) Kaulana, a Events Hooponopono Kui.
• GNU ddrescue Manual ʻenekinia mana (gnu.org/software/ddrescue/manual/ddrescue_manual.html): *Multi-Phase Non-Scraping (-n) a me ke komo pololei (-d) Nā Algorithms Hoʻolaʻa ʻĀpana no ka Haʻalele ʻana i ka Media Magnetic *.
• Kahua ʻike ʻike ʻo Synology (kb.synology.com/en-global/DSM/tutorial/How_can_I_recover_data_from_my_DiskStation_using_a_PC): Ke kau nei mdadm + LVM2 + Btrfs / ext4 Heluhelu-wale na Volumes NAS ma Ubuntu Rescue Systems.
• Hui Pūnaehana ʻOihana Waihona (SNIA): Hōpili ʻIkepili Disk RAID maʻamau (DDF) Mea kikoʻī no ka mea hoʻohana pono ʻo RAID Controller Metadata Portability.