RAID 5, RAID 6, oba ZFS Multiple Drive Drop: Okuzzaawo Synology, QNAP, oba TrueNAS Array eyagwa awatali kusaanyaawo Parity¶

Okuwulira Synology DiskStation, QNAP TS-series, seva ya TrueNAS, oba Linux mdadm seva y'okutereka ekuba enduulu obutasalako nga etegeeza nti Volume Crashed, Pool Degraded / Faulted, oba Multiple Drives Disconnected mbeera ya mangu ey'okutereka ey'emigabo mingi. Drive emu bw’egwa okuva mu a RAID 5, Synology RAID ey'omugatte (SHR-1), oba ZFS RAIDZ1 pool, array etambula ne zero redundancy (degraded state), okuddamu okuzimba bulooka ezibula ku nnyonyi nga osoma buli disiki esigadde mu parallel. Singa drive eyookubiri egwa offline oluvannyuma lw’essaawa, oba okusuula Unrecoverable Read Error (URE) wakati mu kuddamu okuzimba ku disiki ekyusibwa, okubala kwa parity kumenyeka era ekifo ky'okutereka kigenda offline ddala. Abaddukanya abasattira batera okusaanyaawo terabytes za data ezisobola okuddamu mu sikonda nkaaga eziddako nga banyiga Initialize, okuddamu okukola array ne mdadm --create, oba okukaka okuddamu okuzimba amangu ku yintaneeti ku hardware eremereddwa. Singa kaadi yo eya hardware RAID controller yennyini yafa okusinga disiki, kyusa ku Okufa kwa kaadi ya RAID controller, okuyingiza ensengeka y'ebweru, era lwaki tolina kutandika disiki; singa hard drive ssekinnoomu eba enyiga oba egaana okuwuuta, weebuuze hard drive okulemererwa n’okuzzaawo data oba NAS RAID array ekendedde, drive eziwera, n'okuzzaawo ZFS/Btrfs. Singa okulemererwa kuno okutereka kulemesa okukuba ebifaananyi by’abasawo mu ddwaaliro (PACS), okusindika abantu mu bwangu, oba ebikozesebwa ebikuuma obulamu, kuba ku nnamba y’amangu mu kitundu kyo (911 mu Amerika ne Canada, . 112 okwetoloola omukago gwa Bulaaya, . 999 mu Bungereza) oba kebera ku dayirekita yaffe eya ennamba z’amangu.

Kino kikole kati¶

1. YIMIRIZA era TONYIGA "Initialize," "Create Pool," "Repair," oba dduka mdadm --create ku drives ezisuuliddwa: Okunyiga Initialize mu Synology DSM oba QNAP QTS, oba okukola mdadm --create mu Linux, ewandiika superblocks empya ddala era etandika okukuba zero parity stripes mu disiki zonna eza mmemba. Drive esuuliddwa etera okusigala nga ekwata 99.99% wa fayiro zo nga tezifudde, eziteekeddwako akabonero kokka nti tezikwatagana n'omubalirizi w'ebintu byayo.
2. Power down the NAS or server cleanly (oba okusika amaanyi bwe kiba nga kiddamu okuzimba mu bujjuvu okuyita mu drives ezinyiga) okufiriza event counters: Buli sikonda array esigala ku mutimbagano mu mbeera ekendedde ekitundu, enkola y'emirimu eyongera ku superblock Update Time era Events counter ku disiki ezisigaddewo ate disiki esuuliddwa esigala nga efuuse bbugumu ku nnamba y'ekintu ekikadde. Singa NAS egezaako okuddamu okuzimba mu ngeri ey’otoma okubuna drive erimu ebitundu ebibi, okugenda mu maaso n’okusoma drive eyo esobola okuggyako ekizigo kya magnetic platter (head crash).
3. Buli kifo kya drive bay ssaako akabonero oba tape (Bay 1, . Bay 2, . Bay 3, . Bay 4) nga tonnaserengesa tray yonna okufuluma: Nga Linux mdadm, . LVM2, era ZFS track drive order by UUID munda mu buli partition superblock, abafuga hardware enkadde ne manual dmsetup okuddamu okuzimba emmeeza kwetaaga ensengeka entuufu ey’ekifo ekirabika (0, 1, 2, 3...).
4. Togezaako kuzzaawo, kukuŋŋaanya maanyi, oba fsck butereevu ku bivuga bya mmemba ebyasooka; clone buli drive bit-for-bit nga okozesa ddrescue okusooka: Nga tonnadduka mdadm --assemble --force oba zpool import -F, ssaako buli disiki ya mmemba ku siteegi y’emirimu eya Linux ng’oyita mu SATA/SAS obutereevu (si bridge ya USB ya layisi ekweka SMART oba SCT ERC timeouts) era n’ogikolako clone na ddrescue ku sipeeya obulamu oba fayiro z’ebifaananyi.
5. Force-okukuŋŋaanya cloned array strictly mu mbeera ya READ-ONLY (--readonly oba -o readonly=on): Nga bakuŋŋaanya mdadm arrays nga event counters zazo drifted nga zaawukana mu kiseera simultaneous multi-drive drop, bulijjo ziyita --readonly (mdadm --assemble --force --readonly /dev/md0 ...). Mu mbeera y’okusoma kwokka, tewali n’emu mdadm, . LVM2, . ext4, . Btrfs, wadde ZFS asobola okuwandiika ku parity stripes oba okuddamu okuzannya journal esaanyaawo.
6. Koppa data yo esinga obukulu off ku independent external backup drive amangu ddala nga omaze okuteekebwako read-only: Sooka sengula ebiwandiiko byo, databases, virtual machines, n'ebifaananyi ku drive ey'enjawulo, olwo osangule era ogezese disks okuva ku ntandikwa.

Omuti gw'okusalawo: multi-drive RAID ne ZFS pool okusunsula mu bwangu¶

Kozesa flowchart eno okuzuula oba array yo yafuna logical timeout drop (TLER / ERC mismatch), okulemererwa okwa nnamaddala okwa disiki nnyingi mu kitundu ekibi mu kiseera ky'okuddamu okuzimba, oba ekibiina ky'okutunda ekya ZFS (TXG) omusango:

flowchart TD
    Start["RAID 5 / RAID 6 / SHR / ZFS Pool Egudde Oluvannyuma lwa Drives 2+ Okugwa"] --> StopIO["STOP Byonna Rebuilds & Power Down NAS Amangu ddala"]
    StopIO --> Label["Buli Drive Tray ssaako akabonero ku Bay Number (Bay 1 okutuuka ku Bay N)"]
    Label --> ConnectLinux["Yunga Member Drives ku Ubuntu Linux Rescue Host nga oyita mu Direct SATA / HBA"]

    ConnectLinux --> CheckSmart["Kebera SMART & Metadata Okusoma Yokka: smartctl -a / mdadm --okwekenneenya / zdb -l"]
    CheckSmart --> PhysicalHealth{"Waliwo Drives Yonna Enyiga, Egaana Okuwuuta, oba Egwa Offline mu dmesg?"}

    PhysicalHealth -->|"Yee: Okulemererwa kw'ebyuma / omutwe ku 2+ Drives mu RAID 5"| Cleanroom["STOP Okuwa Disiki Amaanyi -> Sindika mu Cleanroom Data Recovery Lab"]
    PhysicalHealth -->|"Nedda: Drives zonna Spin Up (ziyinza okuba ne Pending Bad Sectors oba Timeout Drops)"| CloneAll["Clone Buli Member Drive Nga Okozesa GNU ddrescue (Okuyita mu maaso -> Okusekula emabega)"]

    CloneAll --> DisconnectOrig["Ggyako Original Drives & Kola Strictly ku Cloned Disks / Ebifaananyi"]
    DisconnectOrig --> ArrayType{"Enzimba Ki Ey’okutereka Ekozesebwa?"}

    ArrayType -->|"Synology DSM (SHR / RAID eya bulijjo) ."| SynologyFlow["Kebera Ebibaddewo ng'oyita mu mdadm --examine -> Force-Assemble Okusoma Kwokka -> Kozesa LVM2 / Btrfs Okusoma Kwokka"]
    ArrayType -->|"QNAP QTS (HAL LVM RAID enzito / enzito)"| QnapFlow["Kuŋŋaanya mdadm Okusoma Kwokka -> Kozesa DRBD Layer bwe kiba nga kiriwo -> Sikaani Custom QNAP LVM2 Superblocks"]
    ArrayType -->|"TrueNAS / ZFS (RAIDZ1 / RAIDZ2 / Endabirwamu) ."| ZfsFlow["Dukanya zpool import -d /dev/disk/by-id -> Yingiza Okusoma Kwokka n'Okuzza emabega: zpool import -f -o readonly=ku -F"]
    ArrayType -->|"Linux eya mutindo mdadm"| LinuxFlow["Geraageranya Ebibala Ebibaddewo -> Dukanya mdadm --assemble --force --readonly -> Teeka Enkola ya Fayiro ne -o ro,noload"]

Omutendera 1: Lwaki okukaka okuddamu okuzimba kisaanyaawo data (The URE ne Same-Batch Cascade) .¶

Okutegeera lwaki okunyiga Repair / Rebuild oba okukyusa drive emu n’olinda essaawa 36 kitera okuleeta total array crash, olina okutegeera engeri parity reconstruction gy’essa essira ku mikutu gya magnetic media egyakaddiwa n’ensonga lwaki consumer desktop drives zigwa mu NAS enclosures.

1. Entuufu y’okubala ey’ensobi z’okusoma ezitazuulibwa (UREs) mu kiseera ky'okuddamu okuzimba RAID 5 oba RAIDZ1¶

Mu mmotoka ya 4-drive RAID 5 oba ZFS RAIDZ1 pool, data ne bulooka za parity (P) ziriko emisono ku disiki zonna ennya nga zikozesa OBA ey’enjawulo (XOR) okubala ( ) .D1 XOR D2 XOR D3 = P). Singa Drive 2 efa ddala, array eba ne zero esigaddewo redundancy. Okuddamu okuzimba Drive 2 ku disiki empya ennyo ekyusibwa, NAS erina okusoma buli kitundu kimu okuva ku byte 0 okutuuka ku nkomerero ya disiki ku Drive 1, Drive 3, ne Drive 4 omulundi gumu (Reconstructed D2 = D1 XOR D3 XOR P).

Lowooza ku kiki ekibaawo nga drives ezo ziri hard drive nnya eza terabyte 12 (12,000 gigabyte) ezagulibwa okuva mu batch y’emu ey’okukola emyaka ena egiyise: - Okuddamu okuzimba terabytes 12 kyetaagisa okusoma 36 terabytes (2.88 x 1014 bits ssekinnoomu) obutasalako okumala essaawa 24 ku 48 ku mugugu ogusinga obunene ogw’ebbugumu n’ebyuma. - Hard drive z’abakozesa zitera okuba n’ensobi y’okusoma etasobola kuddamu kuzuulibwa (URE) okunnyonnyola kwa 1 error in 1014 bits (12.5 terabytes), era ebivuga okukaddiwa bikulaakulanya ensobi za silent latent sector (Current_Pending_Sector) mu bitundu ebinyogovu ebya disiki ebitannaba kusomebwa mu myaka. - Akaseera Drive 3 kakuba ekitundu kimu ekitali kya kusoma ekya kilobyte 4 (0.004 megabyte) ekibi nga Drive 2 ekyabula oba ekitundu kyaddamu okuzimbibwa,... XOR equation erina bibiri ebitamanyiddwa (D2 era D3).

2. Lwaki desktop evuga nga temuli TLER / . SCT ERC okugobwa ddala ku sector 1 embi¶

Wano waliwo omutego oguleeta okugwa okw’akatyabaga okwa multi-drive ku Synology, QNAP, ne Linux mdadm:

+-------------------------------------------------------------------------+
| WHY A SINGLE BAD SECTOR DROPS AN ENTIRE DRIVE DURING REBUILD            |
|                                                                         |
| 1. Drive 2 is dead/rebuilding. NAS reads Sector 849,102,400 on Drive 3. |
| 2. Drive 3 hits a weak magnetic spot (Current_Pending_Sector).          |
| 3. Consumer Drive Behavior (No TLER / SCT ERC):                         |
|    - Drive 3 enters deep internal error recovery for 30 to 120 seconds, |
|      stopping all SATA responses while retrying head positioning.       |
| 4. Linux Kernel / NAS Controller Timeout (Default 7 to 30 seconds):     |
|    - Kernel sees Drive 3 unresponsive for >30s -> Issues SATA bus reset |
|    - mdadm assumes Drive 3 is COMPLETELY DEAD and kicks it from array   |
| 5. RESULT: Both Drive 2 and Drive 3 are now marked FAULTY/REMOVED.      |
|    The entire RAID 5 array crashes offline immediately.                 |
+-------------------------------------------------------------------------+

Weetegereze ekyabaddewo ddala: Drive 3 tefudde. Kirina... 99.999999% ebitundu ebiramu n’ebitundu 1 oba 2 byokka ebitali bisomebwa ebya kilobyte 4 (8 kilobytes okuva mu 12 terabytes). Naye kubanga mdadm kicked Drive 3 okuva mu array olw'okuggwaako kw'ekiragiro okwa sikonda 30, volume yonna eraga nga Crashed. Singa okola clone ya Drive 3 ne ddrescue (ekibuuka mu ngeri ey’ekitiibwa kilobytes 8 eza sectors embi n’ekoppa terabytes endala 12) n’okaka okukuŋŋaanya array read-only, ozzaawo kumpi ebitundu 100 ku 100 ebya fayiro zo.

3. Lwaki tolina TEKUdduka mdadm --create oba nyweza "Initialize".¶

Array bw’egwa, enfumo bbiri ezizikiriza zitambula ku mikutu gya yintaneeti: - Enfumo 1: "Ddamu okukola ensengekera n'ensengeka ya drive y'emu ddala ng'okozesa." mdadm --create." ."** Ne bwe kiba nga... --assume-clean, enkyusa ez’omulembe eza mdadm default ku njawulo Data Offset emiwendo okusinziira ku kernel ne mdadm okufulumya (2048 sectors vs 262144 sectors). Ddi lwe mdadm --create awandiika enkyusa empya 1.2 superblock ku offset enkyamu oba nga erina sayizi ya chunk etakwatagana, ewandiika ku LVM2 omutwe gw’eddoboozi ery’omubiri oba ext4 superblock mu maaso g’eddoboozi. - Enfumo 2: "Ggyayo drive eyookubiri eyaakagwa, ogiteekemu, onyige Repair." Singa Drive 1 yavaayo ennaku ssatu emabega (kale data yaayo eri ennaku 3 stale) ate Drive 3 n'egwa eddakiika 10 emabega (kale data yaayo eri mu kiseera kino), okusika n'okuddamu okuyingiza drive enkyamu oba okukaka okuddamu okuzimba nga Drive 1 ewandiikiddwako "active" ne Drive 3 ewandiikiddwako "spare" kireetera NAS overwrite the up-to-date array with 3-day-old stale parity stripes, okwonoona buli dayirekita ne fayiro ebikyusiddwa mu ssaawa 72 eziyise.

Omutendera 2: Okukebera superblocks n’... Event counters safely ku Linux okununula host¶

Okununula Synology, QNAP, TrueNAS, oba Linux array eyagwa mu ngeri ey’obukuumi, ggyawo member drives zonna (oluvannyuma lw’okuwandiika ennamba zaayo eza bay) ozigatte ku desktop PC oba server ekola Ubuntu 22.04 LTS / 24.04 LTS (ne bwe zitandikiddwa okuva ku Live USB stick).

Omutendera 1: Teeka ekitabo ky’okununula abantu mu by’amateeka mu Ubuntu¶

Ggulawo terminal ku kyuma kyo eky'okutaasa ekya Ubuntu era oteekemu ebikozesebwa ebyetaagisa:

sudo apt-get update
sudo apt-get install -y gddrescue smartmontools mdadm lvm2 btrfs-progs xfsprogs zfsutils-linux

Omutendera 2: Maapu ya buli nnukuta ya drive, serial number, n’ensengeka y’okugabanya¶

Wandiika ebyuma byonna ebiyungiddwa ku bulooka n'ennamba zaabyo eza hardware oleme kutabulatabula disiki ya kutaasa na disiki ya NAS ey'ensibuko:

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

Laba emmeeza y'okugabanya eya Synology DSM, QNAP, oba TrueNAS drive eya bulijjo:

Omutendera 3: Kebera mdadm --examine okulaba lwaki ensengekera yagwa (The Event Counter Matrix) .¶

Ku Synology, QNAP, oba Linux mdadm array, okwekenneenya metadata ya RAID ku bitundu bya data y'omukozesa (okugeza, ekitundu 3 ku /dev/sdb3, . /dev/sdc3, . /dev/sdd3, . /dev/sde3) awatali kussaako oba kukyusa kintu kyonna:

sudo mdadm --examine /dev/sd[b-e]3 | grep -E "Filename|Magic|Update Time|Events|Array State|Device Role"

Geraageranya ebifulumizibwa mu ddiivu zonna ennya:

/dev/sdb3:
    Update Time : Sat Oct 31 14:12:09 2026
         Events : 482190
   Device Role : Active device 0
   Array State : A.AA ('A' == active, '.' == missing, 'R' == replacing)

/dev/sdc3:
    Update Time : Thu Oct 29 03:44:11 2026
         Events : 479012
   Device Role : Active device 1
   Array State : AAAA ('A' == active, '.' == missing, 'R' == replacing)

/dev/sdd3:
    Update Time : Sat Oct 31 14:12:04 2026
         Events : 482188
   Device Role : Active device 2
   Array State : A.AA ('A' == active, '.' == missing, 'R' == replacing)

/dev/sde3:
    Update Time : Sat Oct 31 14:12:09 2026
         Events : 482190
   Device Role : Active device 3
   Array State : A.AA ('A' == active, '.' == missing, 'R' == replacing)

Soma bulungi emboozi eno ey’okunoonyereza ku misango: 1. ** /dev/sdc3 (Device Role 1) alina ekintu ekiyitibwa Events okubala kwa 479012 okuva ennaku bbiri eziyise (Oct 29). Drive eyo yasooka kuvaako, ennaku bbiri emabega, era okuva olwo ebadde offline. 2. ** /dev/sdb3 (Role 0) ne ** ./dev/sde3 (Role 3)** bombi balina Events : 482190 ku 14:12:09. 3. ** /dev/sdd3 (Role 2) alina Events : 482188 ku 14:12:04, sekonda 5 zokka n'emikolo 2 emabega sdb3 era sde3. 4. Okumaliriza: Olina okuddamu okuzimba ensengekera ng'okozesa sdb3 (Role 0), . sdd3 (Role 2), era sde3 (Role 3) (nga event counters ze ziwukana ticks 2 zokka), ate nga egenda sdc3 (Role 1) okufuluma ddala.

Omutendera 3: Okukola cloning buli drive bit-for-bit na GNU ddrescue¶

Nga tonnadduka mdadm --assemble --force oba zpool import -F, clone buli drive eyeetaagibwa (naddala drive eyasembayo okuvaamu olw'ebitundu ebibi, nga /dev/sdd mu kyokulabirako waggulu) ku drive empty drive ey’obusobozi obwenkanankana oba obunene, oba mu raw .img fayiro ku voliyumu ennene ey’okutereka.

Omutendera 1: Lwaki ddrescue nga olina mapfile kikakatako¶

Omutindo dd eggyamu embuto ku kitundu ekibi ekisooka, oba ne conv=noerror,sync emala essaawa ng’esika emitwe gyayo egy’okusoma ku luguudo olwonooneddwa ate ng’ejjuza ziro megabytes za data ennungi eziriraanyewo. GNU ddrescue akola mu kuyita okw’amagezi okw’emitendera mingi: 1. Pass 1 (Copying phase): Ekoppa ebitundu byonna ebiramu, ebiddamu amangu ebya disiki ku sipiidi enzijuvu ey’omuddiring’anwa (150 MB/s okutuuka ku 250 MB/s), amangu ago okubuuka ku bulooka zonna ezigenda mpola oba ezitasoma kale drive ereme kubuguma oba okufa nga tonnafuna... 99.99% data ennungi. 2. Pass 2 (Trimming phase): Esoma okutuukira ddala ku mbiriizi entuufu eza buli bulooka embi okuva mu maaso n’emabega (reverse direction) okwawula ekitundu ekyonoonese okutuuka ku kitundu kyennyini ekya kilobyte 4. 3. Pass 3 (Scraping phase): Eddamu okugezaako ebitundu ebyonooneddwa mpola nga tewali kuddamu kuteeka bbaasi mu ngeri eteetaagisa.

Omutendera 2: Dduka emitendera 2 ddrescue ekiragiro kya clone¶

Bulijjo laga fayiro ya maapu (sdd_rescue.map). Singa source drive efa ekutulwako era nga yeetaaga cycle y’amasannyalaze, . ddrescue esoma mapfile n'edda ku byte yennyini we yakoma nga tezzeemu kusoma sector emu eyakoppololwa edda:

# Stage 1: Fast forward pass skipping bad/slow blocks immediately (-n = no-scrape)
sudo ddrescue -f -n -b 4096 /dev/sdd /dev/sdh /root/sdd_rescue.map

# Stage 2: Targeted trimming and 3-retry scrape of the skipped bad sectors
sudo ddrescue -f -d -r 3 -b 4096 /dev/sdd /dev/sdh /root/sdd_rescue.map

Omutendera 4: Okuddamu okuzimba Synology DSM (SHR-1 / . SHR-2 / Classic RAID), QNAP, ne Linux mdadm¶

Oluvannyuma lwa ddiivu zo eza cloned (oba ebyuma bya loopu ebisomebwa byokka) nga byetegefu era nga drive za NAS ezasooka ziggyiddwako bulungi, goberera omutendera gwennyini ogw’okukuŋŋaanya ogutali gwa kusaanyaawo ku pulatifomu yo.

Ensonga A: Linux eya mutindo mdadm oba Synology Classic RAID (RAID 5 / . RAID 6) .¶

1. Mukomye ekitundu kyonna md array Ubuntu gye yandigezezzaako okutandika:

   cat /proc/mdstat
   sudo mdadm --stop /dev/md127
   

2. Force-okukuŋŋaanya ebitundu ebikoleddwa mu ngeri ya cloned (sdb3, . sdd3, . sde3, nga tulekawo ebifudde edda sdc3) mu mu ngeri enkakali ey'okusoma kwokka:

   sudo mdadm --assemble --force --readonly --verbose /dev/md2 /dev/sdb3 /dev/sdd3 /dev/sde3
   

   Laba ebifulumizibwa: mdadm ajja kuwa lipoota forcing event count in /dev/sdd3(2) from 482188 up to 482190 era mdadm: /dev/md2 has been started with 3 drives (out of 4).
3. Kakasa nti array ekola era eriko akabonero (auto-read-only) oba (read-only):

   cat /proc/mdstat
   

Ensonga B: Synology Hybrid RAID (SHR-1 era SHR-2) nga balina LVM2 era Btrfs / . ext4¶

Synology's SHR (Synology Hybrid RAID) ye Linux ya mutindo gwa bitundu 100 ku 100 ** .mdadm + LVM2 (Logical Volume Manager)**: - Drives zonna mu pool ya SHR bweziba nga za sayizi y’emu (okugeza, drives nnya eza terabyte 8), Synology ekola emu mdadm Ekitundu kya RAID 5 (md2 ku kitundu 3 ekya buli drive) era n’ateeka an LVM2 ekibinja ky’obuzito (vg1000 oba vg1) waggulu wa md2. - Bwe ziba nga drives mu SHR pool zirina sayizi ezitabuliddwa (okugeza, drives bbiri eza terabyte 8 ne drives bbiri eza terabyte 16), Synology ekola bbiri ezenjawulo mdadm ensengekera z’ebintu: md2 okubuna Ekitundu 5 (8 TB slice ku drive zonna 4 mu RAID 5) ne md3 okubuna Ekitundu 6 (ekisigadde 8 TB slice ku drives bbiri eza 16 TB mu RAID 1), olwo ne yeegatta md2 + md3 mu kimu LVM2 Ekibinja ky'Omuzingo (vg1000).

Okuteeka voliyumu ya Synology SHR esomebwa yokka ku Ubuntu oluvannyuma lw'okukuŋŋaanya amaanyi /dev/md2 (ne /dev/md3 singa SHR ya sayizi etabuddwamu): 1. Sikaani ku Synology LVM2 Physical Volumes n'ebibinja by'obuzito:

sudo pvscan
sudo vgscan
sudo lvscan

sudo vgchange -ay /dev/vg1000

sudo mkdir -p /mnt/synology_rescue
sudo mount -t ext4 -o ro,noload /dev/vg1000/lv /mnt/synology_rescue

sudo mkdir -p /mnt/synology_rescue
sudo mount -t btrfs -o ro,norecovery /dev/vg1000/lv /mnt/synology_rescue

sudo mount -t btrfs -o ro,norecovery,usebackuproot /dev/vg1000/lv /mnt/synology_rescue

Ensonga C: Synology DSM 7+ Okusiba Volume oba Ensirifu ya Folda egabanyizibwa (eCryptfs) .¶

• Okusiba Folda egabanyizibwa (@sharename@ ebitabo ebirimu ECRYPTFS_FNEK_ENCRYPTED... amannya ga fayiro): Koppa byonna @encrypted_share@ folda off /mnt/synology_rescue sooka, olwo ogiteeke n’ogiggyamu ensirifu ku Ubuntu ng’okozesa ekigambo kyo eky’okusiba ekya Synology oba .key fayiro:

  sudo apt-get install -y ecryptfs-utils
  sudo mkdir -p /mnt/decrypted_share
  sudo mount -t ecryptfs -o ro /mnt/synology_rescue/@encrypted_share@ /mnt/decrypted_share
  

• DSM 7.2+ Ensirifu ya Volume Mu bujjuvu (LUKS): Singa /dev/vg1000/lv alaga TYPE="crypto_LUKS" mu blkid, kisumulule okusoma kwokka ng’okozesa .rkey Fayiro ya Recovery Key eyawanulibwa nga volume ekoleddwa:

  sudo cryptsetup open --readonly --key-file /path/to/Volume1.rkey /dev/vg1000/lv synology_crypt
  sudo mount -o ro,norecovery /dev/mapper/synology_crypt /mnt/synology_rescue
  

Ensonga D: QNAP QTS (LVM2 Thin Provisioning era DRBD layeri) .¶

Yuniti za QNAP NAS ez’omulembe (QTS 4.3+ / 5.x) okuyingizaamu ekitundu ekimu eky’omunda DRBD ekyuma (/dev/drbd1) wakati wa /dev/md1 era nga LVM2 Ekibinja ky'Omuzingo (vg1). 1. Okusooka amaanyi-okukuŋŋaanya ebitundu bya data QNAP (/dev/sd[b-e]3) okusoma-yokka mu /dev/md1:

sudo mdadm --assemble --force --readonly /dev/md1 /dev/sdb3 /dev/sdd3 /dev/sde3

Omutendera 5: Okununula TrueNAS ne Linux ZFS ebidiba (RAIDZ1, . RAIDZ2, . RAIDZ3) .¶

Singa seva yo ey'okutereka ekola TrueNAS CORE, TrueNAS SCALE, Proxmox VE, oba Ubuntu ZFS, okuzzaawo tekukozesa mdadm. Mu kifo ky’ekyo, ZFS egatta omuddukanya eddoboozi n’okukoppa-ku-okuwandiika (CoW) enkola ya fayiro mu nsengeka emu ey'okutunda ezimbiddwa okwetoloola Uberblocks n'ebibinja by'okutunda ebiddiriŋŋana (TXGs)**.

Omutendera 1: Kebera ebiwandiiko 4 ebya ZFS vdev (L0, L1, L2, L3) ku buli disiki eya cloned¶

Buli kitundu kya mmemba wa ZFS kitereka kkopi nnya ezifaanagana ez'ensengeka y'ekidiba n'ensengekera ya Uberblocks 128 ezisembayo (ebiwandiiko bibiri mu maaso g'ekitundu, L0 era L1, n’ebiwandiiko bibiri ku nkomerero yennyini, . L2 era L3). Dduka zdb -l ku buli kitundu ekikoleddwa mu ngeri ya cloned okwekenneenya embeera y’ekidiba n’Ekibinja ky’Enkolagana ekisinga okwewaayo (txg):

sudo zdb -l /dev/sdb2 | grep -E "name:|state:|txg:|guid:" | head -n 20

Geraageranya eby’oku... txg: ennamba mu drive zonna ezikoleddwa mu cloned. Nga bwe kiri mdadm's Events counter, singa disk emu egwa nga wabulayo eddakiika 5 okufiirwa amaanyi oba disiki eyookubiri egwa, its txg ejja kuba wansi katono okusinga disiki ezisigadde.

Omutendera 2: Yingiza ekidiba kya ZFS mu mbeera ya Read-Only enkakali (-o readonly=on) .¶

Sooka okebere pools eziyingizibwa nga tossaako kintu kyonna okuyita mu sudo zpool import -d /dev/disk/by-id. Toyingizangako kifo kya ZFS ekyonoonese oba ekisuuliddwa disiki nnyingi mu mbeera y'okusoma-okuwandiika. Bwe kiyingizibwa mu ggwanga nga... -o readonly=on, ZFS tegezaako kutandikawo resilver, tewandiika Uberblocks mpya, era tesobola kusattira kernel singa maapu y'omu bwengula eyonoonese (spacemap) esangibwa mu kiseera ky'okugabanya okuwandiika:

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

Omutendera 3: Singa zpool import alemererwa ne I/O error oba corrupted data, kozesa TXG Rollback (-F era -T) .¶

Singa drive bbiri zagwa mu sikonda ntono okuva ku ndala mu kiseera ky’okuwandiika okuzitowa okubutuka, Transaction Group eyasembyeyo ennyo (TXG N) eyinza okuba nga tetuukiridde okubuna emisono gya parity egisigaddewo, so nga TXG N-1 oba TXG N-2 (eyawandiikibwa sekondi 5 ku 15 emabega) kirimu ebitundu 100 ku 100 nga tekikyuse munda mu buffer y’empeta ya Uberblock.

1. Gezesa okusooka okusooka okw'okudda emabega okw'okudduka okukalu okutali kwa kuzikiriza (-F -n):

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

   Singa ZFS ekola lipoota Would be able to return pool to its state as of <timestamp>, dduka okuyingiza okuddamu okusomebwa kwokka awatali -n:

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

2. Singa -F nakati tasobola kussaako Uberblocks ezisembyeyo, wandiika ennamba za Uberblock TXG ez’ebyafaayo ne zdb era n’okuyingiza TXG enkadde entongole (-T):

   sudo zdb -e -ul /dev/sdb2 | grep -E "Uberblock|txg =|timestamp ="
   

   Londa a txg ennamba okuva ku sikonda 30 okutuuka ku 60 ng’akabenje kagwawo (okugeza, txg = 9481020) era oyingiza ekidiba ekifumbiddwa mu kiseera ekyo kyennyini mu mbeera y'okusoma kwokka:

   sudo zpool import -f -o readonly=on -R /mnt/zfs_rescue -T 9481020 tank
   

Omutendera 6: Okukakasa oluvannyuma lw’okuzzaawo n’okuziyiza okuddamu okuzimba mu biseera eby’omu maaso¶

Bw’omala okukoppa ebitundu 100 ku 100 ebya data yo gy’ozzeemu okuva /mnt/synology_rescue oba /mnt/zfs_rescue ku kiruubirirwa kya backup ekyetongodde, goberera olukalala luno olw'okukebera hardware n'ensengeka nga tonnazza NAS yo mu mpeereza:

• Togulangako drive za NAS zonna okuva ku musuubuzi omu ku lunaku lumu: Drives okuva mu manufacturing lot emu zigabana head batches ne firmware ezifaanagana, ekizireetera okulemererwa mu ssaawa ntono nga buli emu ezzeemu okuzimba. Tabula batch oba abakola (Seagate IronWolf / Exos, . WD Red Plus / Ultrastar, . Toshiba N300 / MG).
• Kozesa NAS/Enterprise drives zokka nga zirina... TLER / . SCT ERC esobozeseddwa (oba okusitula Linux mdadm SCSI timeouts for desktop drives): Ku ddiivu z'abakozesa mu Linux mdadm, kozesa a udev etteeka okusitula /sys/block/sdX/device/timeout okuva 30 sekondi okutuuka ku 180 sekondi bwe kityo mdadm erinda internal sector remapping mu kifo ky'okusuula drive.
• Migrate arrays nga zirina drives ennene okusinga 8 TB okuva RAID 5 / . SHR-1 / . RAIDZ1 okutuuka ku RAID 6 / . SHR-2 / . RAIDZ2: Nga... 12 TB okutuuka ku 24 TB drives, ebiseera by’okuddamu okuzimba mu parity emu bisukka essaawa 30, ekifuula dual-parity (RAID 6 / . SHR-2 / . RAIDZ2) kikulu okuwangaala ekitundu ekibi mu kiseera ky’okuddamu okukola ffeeza.
• Teekateeka buli mwezi Data Scrubbing (mdadm check / . zpool scrub) ne Extended SMART ebigezo: Okusiimuula buli mwezi kuddaabiriza ensobi mu kitundu ekikusike (Current_Pending_Sector) ate nga drive zonna ziri ku mutimbagano.
• Jjukira nti RAID ya high availability, never a backup: Goberera... 3-2-1 enkola ya backup nga erina kkopi etali ku mutimbagano.

Ennamba z’embeera ez’amangu mu bitundu n’ebiragiro by’ebitongole¶

Okusalako kw’okutereka okukosa emikutu gy’ebyobulamu, empuliziganya ey’amangu, oba ebiwandiiko by’omuntu ebikkirizibwa mu mateeka kyetaagisa okulinnyisibwa mu bwangu:

• Amerika ne Canada: Kuba 911 singa akabenje k’okutereka kalemesa enkola z’obujjanjabi bw’eddwaliro (EHR / PACS) oba okusindika. Lipoota ransomware egendereddwamu NAS (DeadBolt, . LockBit, . Akira) okutuuka ku CISA (1-844-Say-CISA / . cisa.gov/report), Ekitongole kya FBI IC3 (ic3.gov), oba Ekitongole kya Canada ekikola ku by'okwerinda ku mikutu gya yintaneeti (1-833-CYBER-88).
• Bungereza: Kuba 999 ku mbeera ez’amangu ezikosa obukuumi bw’obulamu oba okulabirira okw’amangu okwa NHS. Mu GDPR ya UK, kebera oba okufiirwa ebikwata ku muntu ow’olubeerera nga tekulina buyambi kyetaagisa okutegeeza ICO (0303 123 1113) mu ssaawa 72 zokka.
• Omukago gwa Bulaaya: Kuba 112 okwetoloola EU yonna. Ebitongole ebiri wansi w’ekiragiro kya NIS2 birina okuloopa okuvaako kw’emirimu okunene eri CSIRT yaabwe ey’eggwanga mu ssaawa 24, era okufiirwa kw’ebikwata ku muntu ow’olubeerera kuleeta GDPR Article 33 okukola lipoota mu ssaawa 72.
• Australia ne New Zealand: Kuba 000 mu Australia oba 111 mu New Zealand; loopa ebibaddewo ku bikozesebwa oba ransomware NAS eri ACSC (1300 CYBER1) oba CERT NZ (0800 237 869).
• Buyindi ne Asia-Pacific: Kuba 112 mu Buyindi olw’okukola ku mbeera ez’amangu; okuloopa ebibaddewo mu kutereka ebizimbe eri CERT-In (incident@cert-in.org.in) mu ddirisa ery'essaawa 6 erikakatako.

Ebibuuzo ebitera okubuuzibwa¶

Lwaki yakola mdadm --assemble --force okugaana okutandika RAID 5 array yange na not enough to start the array?¶

Mu mmotoka ya 4-drive RAID 5. Omuntu w’abantu array, waakiri 3 ku 4 ezivuga alina okuyisibwa ku mdadm --okukuŋŋaanya --amaanyi. Dduka sudo mdadm --okwekenneenya / dev / sd [a-z]3 okukebera lwaki drive eyokusatu yagaanibwa: oba oyisizza disiki yonna mu kifo ky'ekitundu (/dev/sdd mu kifo ky’oku... /dev/sdd3), superblock yaayo yateekebwako akabonero nti a Sipeeya wa sipeeya nga okuddamu okuzimba okutaataaganyizibwa, oba auto-assembler ya Ubuntu yasiba drive emu mu inactive /dev/md127 nga bwe kiri ekyuma (dduka sudo mdadm --okuyimiriza --okusika okusooka, oluvannyuma okuddamu okudduka mdadm --okukuŋŋaanya --okukaka --okusoma kwokka).

Watya singa drive etandika okuddamu okuzimba (Recovery = 12%) nga drive eyokubiri tennagwa?¶

Singa Drive 2 yafa, oyingizaamu Drive 2 ekyusiddwa, era ku 12% . rebuild progress Drive 3 evudde ku mutimbagano olw'ekitundu ekibi, tokozesa Drive 2 ekyusibwamu 12%-ezzeemu okuzimbibwa. Wabula, clone the original Okuvuga 3 nga balina drescue (drescue) nga. Olw’okuba Drive 3 yali ya current ddala right up okutuusa okuddamu okuzimba lwe kwayimirira ku 12%, force-assembling the array read-only using Drive 1, the drescue (drescue) nga clone ya Drive 3, ne Drive 4 (okuleka Drive 2 ebweru) ekuwa array ekwatagana.

Nsobola okuteeka drives zange eza Synology oba QNAP mu chassis empya ddala eya NAS singa power supply oba board ya NAS efa?¶

Yee, byombi Synology (Okusenguka kwa HDD) ne QNAP (Okusenguka kwa NAS obutereevu) okutereka ensengeka ya RAID ku drives zennyini. Kyokka, singa drive bbiri oba okusingawo zaali zaava dda (`Volume Crashed`) nga chassis tennafa, tozibuuka munda mu chassis empya eya NAS, kubanga omugezi w'okusenguka ayinza okuwandiika ku bitundutundu oba okutandika okuddamu okuzimba okusaanyaawo. Taasa data esomebwa yokka ku Ubuntu sooka.

Ntegeera ntya oba drive esuuliddwa safe okukola clone nayo ddrescue versus okwetaaga cleanroom lab?¶

Yunga drive ku SATA, gikole amaanyi, era owulirize okumala sekondi 10 ng’olaba sudo dmesg -w. Singa drive ekola amaloboozi aga rhythmic clicking, scraping, oba buzzing, yeekulukuunya waggulu n’ewuuta n’edda wansi, oba etegeeza obusobozi nga 0 B. / . 4.1 GB mu lsblk, emitwe gy’okusoma/okuwandiika giremereddwa mu mubiri: sala amasannyalaze amangu ddala era okozese cleanroom lab. Singa drive ewuuta bulungi, eraga obusobozi bwayo obujjuvu mu lsblk, era nga n’ebiwandiiko byokka Ekitundu_Ekirindirira_Ekiseera kino oba ensobi z'ekiseera ekiwedde, kikole na drescue (drescue) nga.

"RAID 5 Write Hole" kye ki, era kikola kitya ZFS oba Btrfs weewale nga amasannyalaze gavuddeko?¶

Mu nnono RAID 5. Omuntu w’abantu / . RAID 6. Omuntu w’abantu awatali journal, okulongoosa stripe kyetaagisa okuwandiika data block ne parity block ku disks bbiri omulundi gumu. Singa amaanyi galemererwa mid-write, data ne parity tebikyakwatagana (the Wandiika Ekinnya). ZFS (`RAIDZ`) ne Btrfs okumalawo ekituli ky'okuwandiika ng'okozesa Copy-on-Write (`CoW`), okuwandiika buli muguwa omupya mu bulooka ezitakozesebwa n’okulongoosa Uberblock/root pointer yokka oluvannyuma lw’omuguwa omujjuvu okubeera ku disiki.

Nsobola okukozesa dmsetup overlays okugezesa RAID parameters nga tokyusizza cloned drives?¶

Yee, osobola okukola Copy-on-Write (`CoW`) ebibikka waggulu ku byuma bya loopu ebisomebwa byokka (`losetup --read-only`) nga bakozesa Linux dmsetup y'okuteekawo (`snapshot` target) ewagirwa fayiro ezitali nnyingi mu /tmp. Omuntu yenna awandiika okuva mdadm --okukuŋŋaanya --amaanyi oba fsck -y genda mu bikozesebwa omulundi gumu /tmp overlay file, okukuleka okuddamu okuteeka mu second 1 bwe kiba kyetaagisa.

Ebiragiro ebikwatagana nabyo¶

• NAS RAID array ekendedde, drive eziwera, n'okuzzaawo ZFS/Btrfs
• Okufa kwa kaadi ya RAID controller, okuyingiza ensengeka y'ebweru, era lwaki tolina kutandika disiki
• Hard drive okulemererwa n'okuzzaawo data
• SSD ne NVMe mode okusoma kwokka okw'amangu, 0% spare blocks, ne TRIM okulemesa okununula
• rm -rf oba Shift+Delete mu butanwa ku Linux, macOS, oba Windows: okuddamu okuteekebwa mu bwangu okusoma kwokka
• Okutereka kompyuta, enkola ya 3-2-1, n’okuddamu okukola mu mbeera y’akatyabaga
• Ennamba z’amangu okusinziira ku nsi

Ensonda n’obulagirizi obutongole¶

• Linux Raid Wiki (raid.wiki.kernel.org) & Ekitongole kya Linux Kernel: Okuzzaawo RAID ya Sofutiweya Eremereddwa (mdadm --assemble --force --readonly ne Obutakwatagana mu Babala Ebibaddewo).
• Omusingi gwa Sofutiweya ow’obwereere (GNU Project): GNU ddrescue Ekitabo: Okuzzaawo Ebikwata ku Forensic okuva mu Mikutu egyalemererwa.
• Ebiwandiiko bya OpenZFS (openzfs.github.io): Okuzzaawo okuyingiza mu ggwanga mu kidiba kya ZFS (zpool import -F, Ebiyingizibwa mu ggwanga ebisomebwa byokka, ne zdb Okukebera).
• Synology Inc.: Synology Knowledge Center: Nnyinza ntya okukozesa PC okuzzaawo data nga Synology NAS yange ekola bubi (mdadm + LVM2 Okuteeka).
• Ekitongole ky’eggwanga ekikola ku mutindo ne tekinologiya (NIST) (Amerika): Ekiwandiiko eky’enjawulo 800-209: Enkola y’ebyokwerinda ku bikozesebwa mu kutereka.