Unterabschnitte


Software-RAID - GPT-fdisk

Die Voraussetzung

In dieser Anleitung wird 'GPT-fdisk' für die Einteilung der Festplatten benutzt. Mit dem eisfair-Installer 2.2.5 wurde 'GPT-fdisk' als Standard-Partitionierer eingeführt.

Um ein RAID einzurichten, benutzen wir in dieser Beschreibung das Programm 'mdadm'.

Diese Anleitung wurde speziell auf eisfair-Systeme abgestimmt, auf deren Besonderheiten zu gegebener Zeit hingewiesen wird.
Es werden alle Arten von Festplatten (IDE, SATA, SCSI, NVME) unterstützt. Die Festplattenbezeichnung (sd[a-f]) muss jeweils den eigenen Gegebenheiten angepasst werden, unabängig davon, welche Werte in dieser Anleitung verwendet wurden.

eisfair-1  
eisfair-64  
  Das Program 'mdadm' ist im Basis-System enthalten und muss deshalb nicht explizit installiert werden.
   
  Die notwendigen Kernelmodule sind im Kernel integriert und müssen nicht zusätzlich geladen werden.
   
  Wenn das bisherige System auf einer IDE-Festplatte installiert ist, müssen für SCSI- oder SATA-Festplatten noch die entsprechenden Kernelmodule geladen werden.
   
  Durch Ausführen des Befehls 'modprobe $modul' wird das Modul geladen.
   
  ($modul muss in diesem Fall durch den zu ladenden Modulnamen ersetzt werden, z.B. 'ahci')



 eis #  modprobe ahci
 eis #
            


  Sollte der zu ladenden Treiber nicht bekannt sein, dann nutzt man den Befehl
   
  lspci -v
   
  ,um diesen zu ermitteln.

Zusätzliches RAID zu einer Systemplatte einrichten

Ausgangssituation

Ein eisfair-System wurde bereits auf einer IDE-Festplatte installiert und soll um zwei zusätzliche SATA II(I)-Festplatten in einem RAID-1-Verbund ergänzt werden. Damit ergibt sich beispielsweise folgendes Bild:






 eis #  sgdisk -p /dev/hda
 Disk /dev/hda: 625140335 sectors, 298.1 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): C0F8AF45-9DF5-4EFA-8166-7C0156E00E14
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 625140301
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
    1            2048          100351   48.0 MiB    8300  boot
    2          100352          362495   128.0 MiB   8200  swap
    3          362496       625119821   297.9 GiB   8300  root
 eis #
 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): A7B5ED75-DABA-429C-93DF-95896840CE26
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
    1            2048          100351   48.0 MiB    8300  boot
    2          100352          362495   128.0 MiB   8200  swap
    3          362496         2459647   1024.0 MiB  8300  root
    4         2459648      5860512654   2.7 TiB     8300  data
 eis #
 eis # sgdisk -p /dev/sdb 
 Creating new GPT entries.
 Disk /dev/sdb: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 41B3BC0A-F0EF-4D32-AE86-FA54C1921F27
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 5860533101 sectors (2.7 TiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
 eis #
            




Das Basis-System liegt in diesem Beispiel auf der Festplatte /dev/sda.

Die zu einem Software-RAID zusammenzufassenden Festplatten sind

  /dev/sda: wurde schon einmal benutzt und muss nun für
    den Einsatz fit gemacht werden. Erreicht wird dies
    durch das Löschen der vorhandenen Datenstruktur:
  /dev/sdb: hat bisher keine Formatierung - ist also neu



 eis # sgdisk -Z /dev/sda 
 GPT data structures destroyed! You may now partition the disk using fdisk or
 other utilities.
 eis #
            




Beim Ausführen des Befehls 'sgdisk -p DEVICE' ergibt sich somit folgendes Bild:





 eis #  sgdisk -p /dev/hda
 Disk /dev/hda: 625140335 sectors, 298.1 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): C0F8AF45-9DF5-4EFA-8166-7C0156E00E14
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 625140301
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
    1            2048          100351   48.0 MiB    8300  boot
    2          100352          362495   128.0 MiB   8200  swap
    3          362496       625119821   297.9 GiB   8300  root
 eis #
 eis # sgdisk -p /dev/sda 
 Creating new GPT entries.
 Disk /dev/sda: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): A7B5ED75-DABA-429C-93DF-95896840CE26
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
 eis #
 eis # sgdisk -p /dev/sdb 
 Creating new GPT entries.
 Disk /dev/sdb: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 41B3BC0A-F0EF-4D32-AE86-FA54C1921F27
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 5860533101 sectors (2.7 TiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
 eis #
            


Einrichten der ersten Festplatte

Auf der ersten Festplatte wird eine Partition angelegt und deren Typ festgelegt. Mittels des gdisk-Befehls werden folgende Schritte ausgeführt:

  - Partition anlegen (hier wird eine primäre Partition angelegt
    welche die ganze Grösse der Festplatte einnimmt)
  - Typ der Partition auf 'Linux raid auto' ändern
  - Partitionstabelle schreiben





 eis # gdisk /dev/sda 
 GPT fdisk (gdisk) version 1.0.5

 Partition table scan:
   MBR: not present
   BSD: not present
   APM: not present
   GPT: not present

 Creating new GPT entries.

 Command (? for help): p

 Disk /dev/sda: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 43F0A3C8-CF7C-4A44-A753-224B04DBA35A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 5860533101 sectors (2.7 TiB)

 Number  Start (sector)    End (sector)  Size       Code  Name

 Command (? for help): n

 Partition number (1-128, default 1): 1
 First sector (34-5860533134, default = 2048) or {+-}size{KMGTP}: ENTER
 Last sector (2048-5860533134, default = 5860533134) or {+-}size{KMGTP}: ENTER
 Current type is 'Linux filesystem'
 Hex code or GUID (L to show codes, Enter = 8300): L
 0700 Microsoft basic data  0c01 Microsoft reserved    2700 Windows RE
 3000 ONIE boot             3001 ONIE config           4100 PowerPC PReP boot
 4200 Windows LDM data      4201 Windows LDM metadata  7501 IBM GPFS
 7f00 ChromeOS kernel       7f01 ChromeOS root         7f02 ChromeOS reserved
 8200 Linux swap            8300 Linux filesystem      8301 Linux reserved
 8302 Linux /home           8400 Intel Rapid Start     8e00 Linux LVM
 a500 FreeBSD disklabel     a501 FreeBSD boot          a502 FreeBSD swap
 a503 FreeBSD UFS           a504 FreeBSD ZFS           a505 FreeBSD Vinum/RAID
 a580 Midnight BSD data     a581 Midnight BSD boot     a582 Midnight BSD swap
 a583 Midnight BSD UFS      a584 Midnight BSD ZFS      a585 Midnight BSD Vinum
 a800 Apple UFS             a901 NetBSD swap           a902 NetBSD FFS
 a903 NetBSD LFS            a904 NetBSD concatenated   a905 NetBSD encrypted
 a906 NetBSD RAID           ab00 Apple boot            af00 Apple HFS/HFS+
 af01 Apple RAID            af02 Apple RAID offline    af03 Apple label
 af04 AppleTV recovery      af05 Apple Core Storage    be00 Solaris boot
 bf00 Solaris root          bf01 Solaris /usr & Mac Z  bf02 Solaris swap
 bf03 Solaris backup        bf04 Solaris /var          bf05 Solaris /home
 bf06 Solaris alternate se  bf07 Solaris Reserved 1    bf08 Solaris Reserved 2
 bf09 Solaris Reserved 3    bf0a Solaris Reserved 4    bf0b Solaris Reserved 5
 c001 HP-UX data            c002 HP-UX service         ea00 Freedesktop
 eb00 Haiku BFS             ed00 Sony system partitio  ed01 Lenovo system partit
 ef00 EFI System            ef01 MBR partition scheme  ef02 BIOS boot partition
 fb00 VMWare VMFS           fb01 VMWare reserved       fc00 VMWare kcore crash p
 fd00 Linux RAID
 Hex code or GUID (L to show codes, Enter = 8300): fd00
 Changed type of partition to 'Linux RAID'

 Command (? for help): c
 Using 1
 Enter name: data

 Command (? for help): w

 Final checks complete. About to write GPT data. THIS WILL OVERWRITE EXISTING
 PARTITIONS!!

 Do you want to proceed? (Y/N): y
 OK; writing new GUID partition table (GPT) to /dev/sda.
 The operation has completed successfully.
 eis #
            




Für die, die es kurz mögen, die vorherigen Schritte als Einzeiler mit sgdisk:





 eis # sgdisk --new=0:0:-10M -t 1:fd00 -c 1:data /dev/sda 
 Creating new GPT entries.
 Setting name!
 partNum is 0
 REALLY setting name!
 The operation has completed successfully.
 eis #
            


Einrichten der zweiten Festplatte

Nachdem die erste Festplatte komplett eingerichtet wurde, überträgt man deren Partitions-Layout auf die zweite Festplatte und begutachtet das Ergebnis:






 eis # sgdisk -R /dev/hdb /dev/hda 
 The operation has completed successfully. 
 eis #
 eis # sgdisk -G /dev/sdb  
 The operation has completed successfully. 
 eis #
            




 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 625140335 sectors, 298.1 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): C0F8AF45-9DF5-4EFA-8166-7C0156E00E14
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 625140301
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
    1            2048          100351   48.0 MiB    8300  boot
    2          100352          362495   128.0 MiB   8200  swap
    3          362496       625119821   297.9 GiB   8300  root
 eis #
 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 43F0A3C8-CF7C-4A44-A753-224B04DBA35A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 2014 sectors (1007.0 KiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
    1            2048      5860533134   2.7 TiB     FD00  data
 eis #
 eis # sgdisk -p /dev/sdb 
 Disk /dev/sdb: 5860533168 sectors, 2.7 TiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): DC52B1DD-7D2F-4DFD-A783-C1490E4A8D0A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 5860533134
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 2014 sectors (1007.0 KiB)

 Number  Start (sector)    End (sector)  Size       Code  Name
    1            2048      5860533134   2.7 TiB     FD00  data
 eis #
            


Reste eines alten RAID-Verbundes löschen

Falls die Festplatten bereits einmal in einem RAID-Verbund verwendet wurden, muss sichergestellt werden, dass keine Reste der alten RAID-Konfiguration mehr vorhanden sind. Hierzu wird einfach der alte Superblock gelöscht:






 eis # mdadm --zero-superblock /dev/sda1 
 mdadm: Unrecognised md component device - /dev/sda1
 eis #
 eis # mdadm --zero-superblock /dev/sdb1
 mdadm: Unrecognised md component device - /dev/sdb1
 eis #
            





Falls keine Reste eines früheren RAID-Verbundes gefunden werden, wird von mdadm die oben zu sehende Meldung ausgegeben. Das ist an dieser Stelle kein Problem und bei neuen Festplatten absolut normal.

RAID-Verbund anlegen

Die Festplatten sind nun soweit vorbereitet, um sie zu einem RAID-Verbund zusammenzuführen. Der RAID-Verbund wird im nächsten Schritt mit Hilfe von mdadm angelegt:






 eis # mdadm --create /dev/md0 --bitmap=internal --level=1 --raid-disks=2 /dev/sda1 /dev/sdb1 
 Continue creating array? y
 mdadm: Defaulting to version 1.2 metadata
 mdadm: array /dev/md0 started.
 eis #
            


RAID Synchronisation

Jetzt beginnt die Synchronisation der RAID-Festplatten, welche man unbedingt bis zum Ende abwarten sollte. Dieser Vorgang kann, abhängig von der Größe der Festplatten, mehrere Stunden dauern. Den aktuellen Status der Synchronisation kann man sich wie folgt anzeigen lassen:






 eis # cat /proc/mdstat 
 Personalities : [raid1]
 md0 : active raid1 sdc[1] sdb[0]
       2930134272 blocks super 1.2 [2/2] [UU]
       [>....................]  resync =  0.3% (10941248/2930134272) finish=443.2min speed=109772K/sec
       bitmap: 22/22 pages [88KB], 65536KB chunk

 unused devices: <none>
 eis #
            





Beim Aufruf des Befehls 'watch cat /proc/mdstat' erhält man eine alle zwei Sekunden aktualisierte Statusausgabe und kann den Synchronisationsprozess so nebenher im Auge behalten. Zurück auf die Konsole gelangt man mit 'Strg+C':






 eis # watch cat /proc/mdstat 
 Every 2.0s: cat /proc/mdstat                            Sun Jul 20 18:14:42 2014

 Personalities : [raid1]
 md0 : active raid1 sdc[1] sdb[0]
       2930134272 blocks super 1.2 [2/2] [UU]
       [>....................]  resync =  0.3% (10941248/2930134272) finish=443.2min speed=109772K/sec
       bitmap: 22/22 pages [88KB], 65536KB chunk

 unused devices: <none>
 eis #
            


Anlegen des Dateisystems

Nach Abschluss der Synchronisation legt man das Dateisystem an, was üblicherweise einige Minuten dauert. Das Standard-Dateisystem ist 'ext4 - mkfs.ext4', auch wenn es diverse Alternativen gibt. 'Wer die Wahl hat, hat die Qual' ;-)

Zum Einrichten eines 'xfs'-Dateisystems, 'xfs - mkfs.xfs', müssen zusätzlich die 'xfs-tools' von Pack-Eis installiert werden.







 eis # mkfs.ext4 /dev/md0 
 mke2fs 1.42.12 (29-Aug-2014)
 Filesystem label=
 OS type: Linux
 Block size=4096 (log=2)
 Fragment size=4096 (log=2)
 Stride=0 blocks, Stripe width=0 blocks
 183140352 inodes, 732533568 blocks
 36626678 blocks (5.00%) reserved for the super user
 First data block=0
 Maximum filesystem blocks=0
 22356 block groups
 32768 blocks per group, 32768 fragments per group
 8192 inodes per group
 Superblock backups stored on blocks:
         32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,
         4096000, 7962624, 11239424, 20480000, 23887872, 71663616, 78675968,
         102400000, 214990848, 512000000, 550731776, 644972544

 Allocating group tables: done
 Writing inode tables: done
 Creating journal (32768 blocks): done
 Writing superblocks and filesystem accounting information: done
 eis #
            


Bestehendes RAID (auf fremdem System erstellt) integrieren

Dieses Szenario setzt voraus, dass auf einem anderen System ein RAID-Verbund auf einer oder mehreren Platten angelegt wurde und die beinhaltenden Platten in den eisfair-1-Server eingebaut und elektrisch angeschlossen wurden.

RAID starten

Um den existierenden RAID-Verbund mounten zu können, muss dieser zuerst gestartet werden. Dies geht mit folgendem Befehl; ggf. /dev/mdX-Device und die einzelnen Partitionen /dev/sdX bzw. /dev/sdX anpassen.

Beispiel für ein RAID-1 auf zwei Festplatten:






 eis # mdadm --assemble /dev/md0 /dev/sdb1 /dev/sdc1
 mdadm: /dev/md0 has been started with 2 drives (out of 2)
 eis #
            





Beispiel für ein RAID-1 auf zwei Festplatten, wobei gerade nur eine Festplatte angeschlossen ist:






 eis #  mdadm --assemble /dev/md0 /dev/sdb1
 mdadm: /dev/md0 has been started with 1 drive (out of 2)
 eis #
            





Die fehlende Festplatte/Partition kann später mit folgendem Befehl hinzugefügt werden:






 eis # mdadm /dev/md0 -a /dev/sdc1
 eis #
            


RAID mounten

Der RAID-Verbund ist nun eingerichtet und kann gemountet werden. Im Beispiel wird als Mountpoint das Verzeichnis /data/backup verwendet:






 eis # mkdir -p /data/backup
 eis #
 eis # ls -l /data/ 
 total 4
 drwxr-xr-x 2 root root 4096 Jul 20 18:21 backup
 eis #
 eis # mount -t ext4 /dev/md0 /data/backup
 eis #
            





Die neue Partition im RAID-Verbund ist dem System nun bis zum nächsten Reboot bekannt:






 eis # mount 
 /dev/sda3 on / type ext4 (rw,errors=remount-ro)
 /dev/sda1 on /boot type ext4 (rw,errors=remount-ro)
 devpts on /dev/pts type devpts (rw)
 sysfs on /sys type sysfs (rw)
 /dev/md0 on /data/backup type ext4 (rw)
 eis #
            


Systemintegration

Um den RAID-Verbund dauerhaft zu aktivieren, damit er auch einen Neustart übersteht, sind folgende Schritte durchzuführen:

fstab-Datei bearbeiten

Mit Hilfe des Editors 'joe' öffnet man die Datei /etc/fstab und fügt eine zusätzliche Zeile, ebenfalls beginnend mit UUID=, hinzu.

Ermitteln lässt sich der Wert der hinter UUID= erforderlich ist mit foldgendem Kommando.



 eis # blkid -o export /dev/md0 | grep '^UUID'
 eis #
            


Bitte eine Leerzeile am Dateiende nicht vergessen:

  joe /etc/fstab


  UUID=768a... / ext4 defaults,errors=remount-ro 0 1
  UUID=8acf... /boot ext4 defaults,errors=remount-ro 0 1
  UUID=5131... none swap sw 0 0
  /dev/fd0    /media/floppy  auto      defaults,user,noauto     0 0
  /dev/cdrom  /media/cdrom   iso9660   defaults,ro,user,noauto  0 0

  UUID=7g04....    /data/backup   ext4      defaults                 0 0




Initrd bearbeiten

Hinweis
Ab der Kernel Version 6.0.5 linux-4.9.261-eisfair wird dracut als tool zum erzeugen der initrd benutzt


Für die Systeme, denen ein RAID hinzugefügt wurde, ist unter '/usr/share/dracut/config' eine Vorlage abgelegt, 01-dracut-mdraid.conf, die dann nach '/etc/dracut.conf.d' kopiert werden sollte.

Anschliessend wird eine neue initrd mit, mkinitrd erzeugt.

ACHTUNG
Es ist keine weitere behandlung von Hand in der initrd erforderlich. Die Nachfolgende Beschreibung, für die initrd ist veraltet und nur noch für alte Systeme, ohne dracut.


Natürlich ist es auch möglich, die initrd oder initramfs manuell zu bearbeiten und die notwendigen Angaben einzufügen. Dabei gibt es abhängig von der Version des eiskernels verschiedene Vorgehensweisen, da je nach Version mal eine normale initrd und mal eine initramfs verwendet wird.

Vor eiskernel mit der Version 2.11.0 wird eine normale initrd verwendet, die wie folgt bearbeitet wird:






 eis # mkdir -p /tmp/initrd/init
 eis #
 eis # cd /tmp/initrd
 eis #
 eis # cp -a /boot/initrd.gz .
 eis #
 eis # gzip -d initrd.gz
 eis #
 eis # modprobe loop
 eis #
 eis # mount -o loop initrd init
 eis #
            





Es werden der initrd nun das Programm 'mdadm' und die 'mdadm.conf' hinzugefügt:






 eis # cp /sbin/mdadm /tmp/initrd/init/sbin
 eis #
 eis # echo "DEVICE /dev/[sh]d*" >/tmp/initrd/init/etc/mdadm.conf
 eis #
 eis # mdadm --examine --scan >>/tmp/initrd/init/etc/mdadm.conf
 eis #
 eis # echo "/sbin/mdadm --assemble --scan" >>/tmp/initrd/init/linuxrc
 eis #
            





Nun wird die initrd wieder zusammengepackt und zurückkopiert:






 eis # umount init
 eis #
 eis # gzip -9 initrd
 eis #
 eis # cp initrd.gz /boot
 eis #
 eis # reboot
 eis #
            


Ab eiskernel 2.11.0 wird eine initramfs verwendet und das oben beschriebene Verfahren funktioniert so nicht mehr. Die initramfs muss so bearbeitet werden:






 eis # mkdir -p /initramfs
 eis #
 eis # cp /boot/initrd.gz /tmp
 eis #
 eis # gzip -d /tmp/initrd.gz
 eis #
 eis # cd /initramfs
 eis #
 eis # cpio -i < /tmp/initrd
 eis #
            





Es werden der initramfs nun das Programm 'mdadm' und die 'mdadm.conf' hinzugefügt:






 eis # cp /sbin/mdadm /initramfs/sbin
 eis #
 eis # echo "DEVICE /dev/[sh]d*" >/initramfs/etc/mdadm.conf
 eis #
 eis # mdadm --examine --scan >>/initramfs/etc/mdadm.conf
 eis #
            





Nun muss der Datei init, die in der initramfs das Pendant der Datei linuxrc der initrd ist, der Befehl '/sbin/mdadm --assemble --scan' vor den auskommentierten Zeilen hinzugefügt werden. Ab eiskernel 2.18.0 gibt es einen Marker für die genaue Stelle, an der der Befehl eingefügt werden muss. Dieser Marker heisst '# run mdadm here'.







Nun wird die initramfs wieder zusammengepackt und zurückkopiert:






 eis # cd /initramfs
 eis #
 eis # find . | cpio -o -H newc --quiet | gzip -9 > /tmp/initrd.gz
 eis #
 eis # cp /tmp/initrd.gz /boot
 eis #
 eis # reboot
 eis #
            


Reboot und Check

Nach dem Neustart des Servers prüfen wir noch einmal, ob die neue Partition im RAID-Verbund korrekt eingebunden wurde.






 Welcome to eisfair! 
 eisfair-base: 3.4.1
 eiskernel   : 5.15.107-eisfair-1-SMP
 eis #
 eis # mount
 /dev/sda3 on / type ext4 (rw,errors=remount-ro)
 /dev/sda1 on /boot type ext4 (rw,errors=remount-ro)
 devpts on /dev/pts type devpts (rw)
 sysfs on /sys type sysfs (rw)
 /dev/md0 on /data/backup type ext4 (rw)
 eis #
            





An dieser Stelle ist die Konfiguration des RAID-Verbundes und dessen dauerhafte Einbindung in das eisfair-System abgeschlossen.

Den Server komplett auf einem RAID-1 einrichten

Ausgangssituation

Ein eisfair-System wurde bereits mit 4 Partitionen auf einer Festplatte installiert, d.h. es wurde dabei auch eine zusätzliche /data-Partition angelegt. Dieses eisfair-System soll nun komplett auf einen RAID-1-Verbund umziehen. Damit ergibt sich folgendes Bild:

  /dev/sda Erste Festplatte mit dem vorhandenen eisfair-System
  /dev/sda1 /boot
  /dev/sda2 swap-Partition
  /dev/sda3 /
  /dev/sda4 /data
       
  /dev/sdb Zweite 'leere' Festplatte






 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 225A5478-5E44-4F29-A07B-C5237D2C8D0A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
     1            2048          100351   48.0 MiB    8300  boot
     2          100352          362495   128.0 MiB   8200  swap
     3          362496         2459647   1024.0 MiB  8300  root
     4         2459648         8368094   2.8 GiB     8300  data
 eis #
 eis # sgdisk -p /dev/sdb 
 Creating new GPT entries.
 Disk /dev/sdb: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 64A8CE74-7CE2-400A-BCA3-09392C7C8114
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 8388541 sectors (4.0 GiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
 eis #
            


Auf der ersten Festplatte ändert man nun die Partitions-Ids aller vorhandenen Partitionen nach 'fd00' - 'Linux raid autodetect':






 eis #  sgdisk -t 1:fd00 /dev/sda 
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 eis #
 eis #  sgdisk -t 2:fd00 /dev/sda 
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 eis #
 eis #  sgdisk -t 3:fd00 /dev/sda 
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 eis #
 eis #  sgdisk -t 4:fd00 /dev/sda 
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 eis #
            




Für die, die es kurz mögen, gibt es diese Befehle noch einmal als Einzeiler mit sgdisk:





 eis # for i in 1 2 3 4 ; do sgdisk -t ${i}:fd00 /dev/sda ; done 
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 Warning: The kernel is still using the old partition table.
 The new table will be used at the next reboot.
 The operation has completed successfully.
 eis #
            




Jetzt ist ein Reboot erforderlich, da uns die Meldungen darauf hinweisen, dass der Kernel die neue Tabelle erst nach einem Reboot benutzt.

Die Partitionen stellen sich nach dem Reboot wie folgt dar:





 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 225A5478-5E44-4F29-A07B-C5237D2C8D0A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
     1            2048          100351   48.0 MiB    FD00  boot
     2          100352          362495   128.0 MiB   FD00  swap
     3          362496         2459647   1024.0 MiB  FD00  root
     4         2459648         8368094   2.8 GiB     FD00  data
 eis #
            


Einrichten der zweiten Festplatte

Nachdem die erste Festplatte komplett eingerichtet wurde, überträgt man deren Partitions-Layout auf die zweite Festplatte und begutachtet das Ergebnis:





 eis # sgdisk -R /dev/sdb /dev/sda 
 The operation has completed successfully. 
 eis #
 eis # sgdisk -G /dev/sdb  
 The operation has completed successfully. 
 eis #
            






 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 225A5478-5E44-4F29-A07B-C5237D2C8D0A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
     1            2048          100351   48.0 MiB    FD00  boot
     2          100352          362495   128.0 MiB   FD00  swap
     3          362496         2459647   1024.0 MiB  FD00  root
     4         2459648         8368094   2.8 GiB     FD00  data
 eis #
 eis # sgdisk -p /dev/sdb 
 Disk /dev/sdb: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): D2ADEFB9-7DD6-4942-BB4B-3EAAF84DA137
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
     1            2048          100351   48.0 MiB    FD00  boot
     2          100352          362495   128.0 MiB   FD00  swap
     3          362496         2459647   1024.0 MiB  FD00  root
     4         2459648         8368094   2.8 GiB     FD00  data
 eis #
            





Reste eines alten RAID-Verbundes löschen

Falls die Festplatten bereits einmal in einem RAID-Verbund verwendet wurden, muss sichergestellt werden, dass keine Reste der alten RAID-Konfiguration mehr vorhanden sind. Hierzu wird einfach der alte Superblock gelöscht:






 eis # mdadm --zero-superblock /dev/sdb1 
 mdadm: Unrecognised md component device - /dev/sdb1
 eis #
 eis # mdadm --zero-superblock /dev/sdb2
 mdadm: Unrecognised md component device - /dev/sdb2
 eis #
 eis # mdadm --zero-superblock /dev/sdb3 
 mdadm: Unrecognised md component device - /dev/sdb3
 eis #
 eis # mdadm --zero-superblock /dev/sdb4
 mdadm: Unrecognised md component device - /dev/sdb4
 eis #
            





Falls keine Reste eines früheren RAID-Verbundes gefunden werden, wird die oben zu sehende Meldung ausgegeben. Das ist an dieser Stelle kein Problem und bei neuen Festplatten absolut normal.




RAID-Verbund anlegen

Die Festplatten sind nun soweit vorbereitet, dass ein RAID-Verbund angelegt werden kann. Hierzu verwendet man mdadm. Zur besseren Orientierung wird empfohlen, die Bezeichnungen der 'md'-Devices an die der vorherigen Partitionen anzulehnen:

  /dev/sda1 /dev/md1
  /dev/sda2 /dev/md2
  /dev/sda3 /dev/md3
  /dev/sda4 /dev/md4
       

Da die erste Festplatte die zu kopierenden Daten enthält, wird sie beim Anlegen des RAID-Verbundes als 'missing' deklariert. Das Schlüsselwort 'missing' sagt lediglich aus, dass noch Devices fehlen und bezieht sich nicht auf das danach angegebene Device.

Es ist unbedingt darauf zu achten, dass die '/boot'-Partition, also /dev/md1, mit '--metadata=0.90' angelegt wird, da der Kernel nur davon booten kann.






 eis # mdadm --create /dev/md1 --bitmap=internal --level=1 --metadata=0.90 --raid-disks=2 missing /dev/sdb1 
 mdadm: array /dev/md1 started.
 eis #
 eis # mdadm --create /dev/md2 --bitmap=internal --level=1 --raid-disks=2 missing /dev/sdb2 
 mdadm: Note: this array has metadata at the start and
     may not be suitable as a boot device.  If you plan to
     store '/boot' on this device please ensure that
     your boot-loader understands md/v1.x metadata, or use
     --metadata=0.90
 Continue creating array? y 
 mdadm: Defaulting to version 1.2 metadata
 mdadm: array /dev/md2 started.
 eis #
 eis # mdadm --create /dev/md3 --bitmap=internal --level=1 --raid-disks=2 missing /dev/sdb3 
  mdadm: Note: this array has metadata at the start and
     may not be suitable as a boot device.  If you plan to
     store '/boot' on this device please ensure that
     your boot-loader understands md/v1.x metadata, or use
     --metadata=0.90
 Continue creating array? y 
 mdadm: Defaulting to version 1.2 metadata
 mdadm: array /dev/md3 started.
 eis #
 eis # mdadm --create /dev/md4 --bitmap=internal --level=1 --raid-disks=2 missing /dev/sdb4 
  mdadm: Note: this array has metadata at the start and
     may not be suitable as a boot device.  If you plan to
     store '/boot' on this device please ensure that
     your boot-loader understands md/v1.x metadata, or use
     --metadata=0.90
 Continue creating array? y 
 mdadm: Defaulting to version 1.2 metadata
 mdadm: array /dev/md4 started.
 eis #
            


Das Ergebnis lässt man sich zur Kontrolle anzeigen:






 eis # cat /proc/mdstat 
 Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] [multipath]
 md4 : active raid1 sdb4[1]
        2952128 blocks super 1.2 [2/1] [_U]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 md3 : active raid1 sdb3[1]
        1048000 blocks super 1.2 [2/1] [_U]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 md2 : active raid1 sdb2[1]
        130944 blocks super 1.2 [2/1] [_U]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 md1 : active raid1 sdb1[1]
        49088 blocks [2/1] [_U]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 unused devices: <none>
 eis #
            





Anlegen des Dateisystems

Nun legt man das Dateisystem an, was üblicherweise einige Minuten dauert. Das Standard-Dateisystem vieler Linux-Systeme ist auch heute noch 'ext3 - mkfs.ext3', auch wenn es diverse Alternativen gibt. 'Wer die Wahl hat, hat die Qual' ;-)

Neu für eisfair hinzugekommene Dateisysteme sind 'ext4 - mkfs.ext4' und 'xfs - mkfs.xfs'. Zum Einrichten eines 'xfs'-Dateisystems müssen zusätzlich die 'xfs-tools' von Pack-Eis installiert werden.




  mkfs.ext4 /dev/md1
  mkfs.ext4 /dev/md3
  mkfs.ext4 /dev/md4
  mkswap /dev/md2






 eis # mkfs.etx4 /dev/md1 
 mke2fs 1.42.12 (29-Aug-2014)
 Filesystem label=
 OS type: Linux
 Block size=1024 (log=0)
 Fragment size=1024 (log=0)
 Stride=0 blocks, Stripe width=0 blocks
 12288 inodes, 49088 blocks
 2454 blocks (5.00%) reserved for the super user
 First data block=1
 Maximum filesystem blocks=50331648
 6 block groups
 8192 blocks per group, 8192 fragments per group
 2048 inodes per group
 Superblock backups stored on blocks:
          8193, 24577, 40961

 Allocating group tables: done
 Writing inode tables: done
 Creating journal (4096 blocks): done
 Writing superblocks and filesystem accounting information: done
 eis #
 eis # mkfs.etx4 /dev/md3 
 mke2fs 1.42.12 (29-Aug-2014)
 Filesystem label=
 OS type: Linux
 Block size=4096 (log=2)
 Fragment size=4096 (log=2)
 Stride=0 blocks, Stripe width=0 blocks
 65536 inodes, 262000 blocks
 13100 blocks (5.00%) reserved for the super user
 First data block=0
 Maximum filesystem blocks=268435456
 8 block groups
 32768 blocks per group, 32768 fragments per group
 8192 inodes per group
 Superblock backups stored on blocks:
          32768, 98304, 163840, 229376

 Allocating group tables: done
 Writing inode tables: done
 Creating journal (4096 blocks): done
 Writing superblocks and filesystem accounting information: done
 eis #
 eis # mkfs.etx4 /dev/md4 
 mke2fs 1.42.12 (29-Aug-2014)
 Filesystem label=
 OS type: Linux
 Block size=4096 (log=2)
 Fragment size=4096 (log=2)
 Stride=0 blocks, Stripe width=0 blocks
 184736 inodes, 738032 blocks
 36901 blocks (5.00%) reserved for the super user
 First data block=0
 Maximum filesystem blocks=759169024
 23 block groups
 32768 blocks per group, 32768 fragments per group
 8032 inodes per group
 Superblock backups stored on blocks:
          32768, 98304, 163840, 229376, 294912

 Allocating group tables: done
 Writing inode tables: done
 Creating journal (16384 blocks): done
 Writing superblocks and filesystem accounting information: done
 eis #
 eis # mkswap /dev/md2 
 Setting up swapspace version 1, size = 130940 KiB
 no label, UUID=30231248-4bf2-40c9-a09c-d69cc4ffc53a
 eis #
            







Einbinden der RAID-Festplatte

Das Root-Device (/dev/md3) mountet man nach /mnt und legt dort ein neues boot-Verzeichnis an. Dieses boot-Verzeichnis verwendet man nun als Mountpoint für das Boot-Device (/dev/md1):






 eis # mount -t ext4 /dev/md3 /mnt
 eis #
 eis # mkdir -p /mnt/boot
 eis #
 eis # mount -t ext4 /dev/md1 /mnt/boot
 eis #
            





Übertragen der Daten

Die Partitionen mit dem Root-Verzeichnis sowie dem Boot-Verzeichnis werden auf die neu erstellten Multiple-Devices kopiert.

Da der Kopiervorgang selbst einige Minuten dauern kann, wartet man so lange, bis der Eingabe-Prompt wieder erscheint:






 eis #  cd /
 eis #
 eis #  find . -xdev | cpio -pmv /mnt
 eis #
 eis #  cd /boot
 eis #
 eis #  find . -xdev | cpio -pmv /mnt/boot
 eis #
            


ACHTUNG
Änderung ab base 2.7.7 mit UDEV


Im Anschluss müssen nun noch im /dev Verzeichnis die Devices console und null angelegt werden. Diese Devices müssen beim boot vorhanden sein.

Dazu wechselt man in das Verzeichnis:

cd /mnt/dev

mknod -m 0660 console c 5 1
mknod -m 0660 null c 1 3

fstab-Datei bearbeiten

Mit Hilfe des Editors 'joe' öffnet man die Datei /mnt/etc/fstab und passt sie an die neuen Multiple-Device-Namen an. Bitte eine Leerzeile am Dateiende nicht vergessen:

Ermitteln lässt sich der Wert der jeweils hinter UUID= erforderlich ist mit foldgendem Kommando. Bitte den Wert /dev/mdX bitte jeweils anpassen.



 eis # blkid -o export /dev/md1 | grep '^UUID'
 eis #
            


  joe /mnt/etc/fstab


  UUID=768a... / ext4 defaults,errors=remount-ro 0 1
  UUID=8acf... /boot ext4 defaults,errors=remount-ro 0 1
  UUID=5131... none swap sw 0 0
  UUID=7g04... /data ext4 defaults 0 0
   /dev/fd0    /media/floppy  auto      defaults,user,noauto     0 0
  /dev/cdrom  /media/cdrom   iso9660   defaults,ro,user,noauto  0 0




Initrd bearbeiten

Hinweis
Ab der Kernel Version 6.0.5 linux-4.9.261-eisfair wird dracut als tool zum erzeugen der initrd benutzt


Für die Systeme, denen ein RAID hinzugefügt wurde, ist unter '/usr/share/dracut/config' eine Vorlage abgelegt, 01-dracut-mdraid.conf, die dann nach '/etc/dracut.conf.d' kopiert werden sollte.

Anschliessend wird eine neue initrd mit, mkinitrd erzeugt.

ACHTUNG
Es ist keine weitere behandlung von Hand in der initrd erforderlich. Die Nachfolgende Beschreibung, für die initrd ist veraltet und nur noch für alte Systeme, ohne dracut.


Damit das zukünftige RAID auch startet, muss initrd bzw. initramfs geöffnet und manipuliert werden. Dabei gibt es abhängig von der Version des eiskernels verschiedene Vorgehensweisen, da je nach Version mal eine normale initrd und mal eine initramfs verwendet wird.

Vor eiskernel mit der Version 2.11.0 wird eine normale initrd verwendet, die wie folgt bearbeitet wird:






 eis # mkdir -p /tmp/initrd/init
 eis #
 eis # cd /tmp/initrd
 eis #
 eis # cp -a /mnt/boot/initrd.gz .
 eis #
 eis # gzip -d initrd.gz
 eis #
 eis # modprobe loop
 eis #
 eis # mount -o loop initrd init
 eis #
            


Es werden nun das Programm 'mdadm' und die 'mdadm.conf' der initrd hinzugefügt:





 eis # cp /sbin/mdadm /tmp/initrd/init/sbin
 eis #
 eis # echo "DEVICE /dev/[sh]d*" >/tmp/initrd/init/etc/mdadm.conf
 eis #
 eis # mdadm --examine --scan >>/tmp/initrd/init/etc/mdadm.conf
 eis #
 eis # echo "/sbin/mdadm --assemble --scan" >>/tmp/initrd/init/linuxrc
 eis #
            




Nun wird die initrd wieder zusammengepackt und zurückkopiert:






 eis # umount init
 eis #
 eis # gzip -9 initrd
 eis #
 eis # cp initrd.gz /mnt/boot
 eis #
            


Ab eiskernel 2.11.0 wird eine initramfs verwendet und das oben beschriebene Verfahren funktioniert so nicht mehr. Die initramfs muss so bearbeitet werden:






 eis # mkdir -p /initramfs
 eis #
 eis # cp /mnt/boot/initrd.gz /tmp
 eis #
 eis # gzip -d /tmp/initrd.gz
 eis #
 eis # cd /initramfs
 eis #
 eis # cpio -i < /tmp/initrd
 eis #
            





Es werden der initramfs nun das Programm 'mdadm' und die 'mdadm.conf' hinzugefügt:






 eis # cp /sbin/mdadm /initramfs/sbin
 eis #
 eis # echo "DEVICE /dev/[sh]d*" >/initramfs/etc/mdadm.conf
 eis #
 eis # mdadm --examine --scan >>/initramfs/etc/mdadm.conf
 eis #
            





Nun muss der Datei init, die in der initramfs das Pendant der Datei linuxrc der initrd ist, der Befehl '/sbin/mdadm --assemble --scan' vor den auskommentierten Zeilen hinzugefügt werden. Ab eiskernel 2.18.0 gibt es einen Marker für die genaue Stelle, an der der Befehl eingefügt werden muss. Dieser Marker heisst '# run mdadm here'.







Nun wird die initramfs wieder zusammengepackt und zurückkopiert:






 eis # cd /initramfs
 eis #
 eis # find . | cpio -o -H newc --quiet | gzip -9 > /tmp/initrd.gz
 eis #
 eis # cp /tmp/initrd.gz /mnt/boot
 eis #
            








RAID zusammenführen

Wenn alle bisherigen Schritte ohne Probleme durchgeführt wurden, läuft der RAID-Verbund zwar bereits, jedoch ohne aktive Redundanz, da die Festplatte /dev/sda noch nicht Teil des RAID-Verbundes ist:






 eis # mount 
 /dev/md3 on / type ext4 (rw,errors=remount-ro)
 /dev/md1 on /boot type ext4 (rw,errors=remount-ro)
 /dev/md4 on /data type ext4 (rw)
 devpts on /dev/pts type devpts (rw)
 sysfs on /sys type sysfs (rw)
 eis #
            





Die Festplatte /dev/sda wird nun in den RAID-Verbund integriert:






 eis # mdadm /dev/md1 --add  /dev/sda1 
 mdadm: added /dev/sda1
 eis #
 eis # mdadm /dev/md2 --add  /dev/sda2 
 mdadm: added /dev/sda2
 eis #
 eis # mdadm /dev/md3 --add  /dev/sda3 
 mdadm: added /dev/sda3
 eis #
 eis # mdadm /dev/md4 --add  /dev/sda4 
 mdadm: added /dev/sda4
 eis #
            


RAID-Synchronisation

Jetzt beginnt die Synchronisation der RAID-Festplatten, welche man unbedingt bis zum Ende abwarten sollte. Dieser Vorgang kann, abhängig von der Größe der Festplatten, mehrere Stunden dauern. Den aktuellen Status der Synchronisation kann man sich wie folgt anzeigen lassen:



 eis # cat /proc/mdstat 
 Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] [multipath]
 md2 : active raid1 sda2[0] sdb2[1]
        130944 blocks super 1.2 [2/2] [UU]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 md3 : active raid1 sda3[2] sdb3[1]
        1048000 blocks super 1.2 [2/1] [_U]
        [========>............]  recovery = 40.9% (429504/1048000)
 finish=0.3min speed=28633K/sec
        bitmap: 1/1 pages [4KB], 65536KB chunk

 md4 : active raid1 sda4[2] sdb4[1]
        2952128 blocks super 1.2 [2/1] [_U]
          resync=DELAYED
        bitmap: 1/1 pages [4KB], 65536KB chunk

 md1 : active raid1 sda1[0] sdb1[1]
        49088 blocks [2/2] [UU]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 unused devices: <none>
 eis #
            




Beim Aufruf des Befehls 'watch cat /proc/mdstat' erhält man eine alle zwei Sekunden aktualisierte Statusausgabe und kann den Synchronisationsprozess so nebenher im Auge behalten. Zurück auf die Konsole gelangt man mit 'Strg+C':



 eis # watch cat /proc/mdstat 
 Every 2.0s: cat /proc/mdstat                            Thu Sep  8 17:44:42 2011

 Personalities : [linear] [raid0] [raid1] [raid10] [raid6] [raid5] [raid4] [multipath]
 md2 : active raid1 sda2[0] sdb2[1]
        130944 blocks super 1.2 [2/2] [UU]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 md3 : active raid1 sda3[2] sdb3[1]
        1048000 blocks super 1.2 [2/1] [_U]
        [========>............]  recovery = 40.9% (429504/1048000)
 finish=0.3min speed=28633K/sec
        bitmap: 1/1 pages [4KB], 65536KB chunk

 md4 : active raid1 sda4[2] sdb4[1]
        2952128 blocks super 1.2 [2/1] [_U]
          resync=DELAYED
        bitmap: 1/1 pages [4KB], 65536KB chunk

 md1 : active raid1 sda1[0] sdb1[1]
        49088 blocks [2/2] [UU]
        bitmap: 0/1 pages [0KB], 65536KB chunk

 unused devices: <none>
 eis #
            


Abschlußarbeiten



Reboot und Check

An dieser Stelle ist die Konfiguration des RAID-Verbundes und dessen dauerhafte Einbindung in das eisfair-System abgeschlossen. Dies sollte noch durch einen Neustart des Servers überprüft werden:



 eis # Welcome to eisfair! 
 eisfair-base: 3.4.1
 eiskernel   : 5.15.107-eisfair-1-SMP
 eis #
 eis # mount

 /dev/md3 on / type ext4 (rw,errors=remount-ro)
 /dev/md1 on /boot type ext4 (rw,errors=remount-ro)
 /dev/md4 on /data type ext4 (rw)
 devpts on /dev/pts type devpts (rw)
 sysfs on /sys type sysfs (rw)
 eis #
            





Den Server komplett auf einem RAID-5 einrichten




Die Installation eines RAID-5 erfolgt analog eines RAID-1 mit einer dritten Festplatte.

Der wesentlichste und wichtigste Unterschied ist, dass ein Linux-System nicht von einem Software-RAID-5 booten kann. Das heisst nun aber nicht, dass sich ein System mit Software-RAID-5 nicht einrichten und betreiben lässt. Es benötigt dafür nur eine Boot-Partition, die kein RAID-5 enthält. Dafür ist ein RAID-1 genau richtig.




Ein Software-RAID wird vom Kernel bereit gestellt. Das bedeutet, der Kernel muss selbst erst einmal geladen sein, um dieses Software-RAID zu initalisieren. Wenn nun /boot mit dem zu ladenen Kernel aber auf einem RAID-5 liegen würde, bei dem die Daten über 3 Festplatten verteilt sind, müßte der Kernel von einem Array geladen werden, welches nur der Kernel selber initialisieren kann ...

Test

Weitere Tests lassen sich realisieren, indem man einzelne Festplatten nacheinander hardwareseitig abklemmt. Es ist darauf zu achten, dass die Synchronisation immer vollständig erfolgt ist, bevor ein Reboot ausgeführt wird. Ist alles korrekt konfiguriert, muss das System von jedem Laufwerk ohne Fehlermeldungen hochfahren.

Desaster recovery

Bei Ausfall einer Festplatte in einem Software-RAID-Verbund sind im Wesentlichen die gleichen Schritte wie bei dem Hinzufügen der 2. Festplatte, weiter vorne beschrieben, auszuführen.






 eis # hdparm -i /dev/sdb 

 /dev/sdb:

  Model=WDC WD10EARS-00Y5B1, FwRev=80.00A80, SerialNo=WD-WCAV5F175350
  Config={ HardSect NotMFM HdSw>15uSec SpinMotCtl Fixed DTR>5Mbs FmtGapReq }
  RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=50
  BuffType=unknown, BuffSize=8192kB, MaxMultSect=16, MultSect=16
  CurCHS=65535/1/63, CurSects=4128705, LBA=yes, LBAsects=625142448
  IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
  PIO modes:  pio0 pio3 pio4
  DMA modes:  mdma0 mdma1 mdma2
  UDMA modes: udma0 udma1 udma2 udma3 udma4 udma5 *udma6
  AdvancedPM=no WriteCache=enabled
  Drive conforms to: Unspecified:  ATA/ATAPI-1,2,3,4,5,6,7

  * signifies the current active mode
 eis #
            










 eis # mdadm /dev/md1 --fail /dev/sdb1 
 mdadm: set /dev/sdb1 faulty in /dev/md1
 eis #
 eis # mdadm /dev/md2 --fail /dev/sdb2 
 mdadm: set /dev/sdb2 faulty in /dev/md2
 eis #
 eis # mdadm /dev/md3 --fail /dev/sdb3 
 mdadm: set /dev/sdb3 faulty in /dev/md3
 eis #
 eis # mdadm /dev/md4 --fail /dev/sdb4 
 mdadm: set /dev/sdb4 faulty in /dev/md4
 eis #
            










 eis # mdadm /dev/md1 --remove /dev/sdb1 
 mdadm: hot removed /dev/hsdb1
 eis #
 eis # mdadm /dev/md2 --remove /dev/sdb2 
 mdadm: hot removed /dev/sdb2
 eis #
 eis # mdadm /dev/md3 --remove /dev/sdb3 
 mdadm: hot removed /dev/sdb3
 eis #
 eis # mdadm /dev/md4 --remove /dev/sdb4 
 mdadm: hot removed /dev/sdb4
 eis #
            










 eis # sgdisk -R /dev/sdb /dev/sda 
 The operation has completed successfully. 
 eis #
 eis # sgdisk -G /dev/sdb  
 The operation has completed successfully. 
 eis #
            







 eis # sgdisk -p /dev/sda 
 Disk /dev/sda: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): 225A5478-5E44-4F29-A07B-C5237D2C8D0A
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
     1            2048          100351   48.0 MiB    FD00  boot
     2          100352          362495   128.0 MiB   FD00  swap
     3          362496         2459647   1024.0 MiB  FD00  root
     4         2459648         8368094   2.8 GiB     FD00  data
 eis #
 eis # sgdisk -p /dev/sdb 
 Disk /dev/sdb: 8388608 sectors, 4.0 GiB
 Logical sector size: 512 bytes
 Disk identifier (GUID): D2ADEFB9-7DD6-4942-BB4B-3EAAF84DA137
 Partition table holds up to 128 entries
 First usable sector is 34, last usable sector is 8388574
 Partitions will be aligned on 2048-sector boundaries
 Total free space is 22494 sectors (11.0 MiB)

 Number  Start (sector)    End (sector)  Size        Code  Name
     1            2048          100351   48.0 MiB    FD00  boot
     2          100352          362495   128.0 MiB   FD00  swap
     3          362496         2459647   1024.0 MiB  FD00  root
     4         2459648         8368094   2.8 GiB     FD00  data
 eis #
            










 eis # mdadm /dev/md1 --add  /dev/sdb1 
 mdadm: added /dev/sdb1
 eis #
 eis # mdadm /dev/md2 --add  /dev/sdb2 
 mdadm: added /dev/sdb2
 eis #
 eis # mdadm /dev/md3 --add  /dev/sdb3 
 mdadm: added /dev/sdb3
 eis #
 eis # mdadm /dev/md4 --add  /dev/sdb4 
 mdadm: added /dev/sdb4
 eis #
            




Nach der Synchronisation ist der RAID-Verbund wieder vollständig.

Endbemerkungen

Es ist unbedingt zu beachten, dass ein RAID-Verbund kein Backup ersetzt. Aus diesem Grund wird empfohlen, eine verlässliche Sicherungsstrategie zu erstellen.
Darüberhinaus sollte vor der Überführung eines bestehenden eisfair-Systems in einen RAID-Verbund, wie vor allen großen Änderungen an der Hard- oder Software, eine Datensicherung auf einen externen Datenträger durchgeführt werden.
Für die Erstellung dieser Anleitung wurde u.a. auf die auf eisfair.org abgelegten Anleitungen von Jürgen Witt und Manfred Reiland zurückgegriffen.