Difference between revisions of "Repairing RAID of the swap"

From SME Server
Jump to navigationJump to search
(Created page with "== Repairing manually the RAID of the swap == Author: Arnaud === Requirements: === {{Warning box|'''This howto works for SME9.1, raid1, nolvm. Some adaptations may be necess...")
 
Line 4: Line 4:
  
 
=== Requirements: ===
 
=== Requirements: ===
{{Warning box|'''This howto works for SME9.1, raid1, nolvm.
+
{{Warning box|'''This howto works for SME9.1, raid1, nolvm and only for the Raid-device concerning the swap.
 
Some adaptations may be necessary for other version of SME or for other parameters concerning the RAID and the LVM! '''}}
 
Some adaptations may be necessary for other version of SME or for other parameters concerning the RAID and the LVM! '''}}
  
=== Starting situation: ===
+
Because the SME is able to run without swap, the job can be done directly from the running SME, without any LiveCD or Rescue-mode.
 +
 
 +
Adapt the name of the partitions (hdX. sdX etc…) to your case.
 +
 
 +
=== The starting point: the device for the swap can't get sync ===
 +
This can occur when a new disk has been added to the SME and that this disk is "a little bit" smaller than the disk what is already running.
 +
The raid sync (e.g. started from the console) works of "/", for "/boot" but not for the swap because os missing some space on the added disk.
 +
 
 +
==== look to the current state of the RAID: ====
 +
 
 +
    # cat /proc/mdstat
 +
    Personalities : [raid1]
 +
    md0 : active raid1 vda1[0] vdb1[2]
 +
          255936 blocks super 1.0 [2/2] [UU]
 +
         
 +
    md2 : active raid1 vda3[0]
 +
          2095104 blocks super 1.1 [2/1] [U_]
 +
         
 +
    md1 : active raid1 vda2[0] vdb3[2]
 +
          18600960 blocks super 1.1 [2/2] [UU]
 +
          bitmap: 0/1 pages [0KB], 65536KB chunk
 +
    unused devices: <none>
 +
 
 +
As indicated over the console too, md2 runs with only 1 disk (vda3). The disk vdb2 is missing into the RAID.
 +
The reason is:
 +
    # mdadm --manage /dev/md2 --add /dev/vdb2
 +
    mdadm: /dev/vdb2 not large enough to join array (the disk#2 has been added to the machine afterwards)

Revision as of 20:51, 10 February 2016

Repairing manually the RAID of the swap

Author: Arnaud

Requirements:

Warning.png Warning:
This howto works for SME9.1, raid1, nolvm and only for the Raid-device concerning the swap.

Some adaptations may be necessary for other version of SME or for other parameters concerning the RAID and the LVM!


Because the SME is able to run without swap, the job can be done directly from the running SME, without any LiveCD or Rescue-mode.

Adapt the name of the partitions (hdX. sdX etc…) to your case.

The starting point: the device for the swap can't get sync

This can occur when a new disk has been added to the SME and that this disk is "a little bit" smaller than the disk what is already running. The raid sync (e.g. started from the console) works of "/", for "/boot" but not for the swap because os missing some space on the added disk.

look to the current state of the RAID:

   # cat /proc/mdstat
   Personalities : [raid1] 
   md0 : active raid1 vda1[0] vdb1[2]
         255936 blocks super 1.0 [2/2] [UU]
         
   md2 : active raid1 vda3[0]
         2095104 blocks super 1.1 [2/1] [U_]
         
   md1 : active raid1 vda2[0] vdb3[2]
         18600960 blocks super 1.1 [2/2] [UU]
         bitmap: 0/1 pages [0KB], 65536KB chunk
   unused devices: <none> 

As indicated over the console too, md2 runs with only 1 disk (vda3). The disk vdb2 is missing into the RAID. The reason is:

    # mdadm --manage /dev/md2 --add /dev/vdb2
   mdadm: /dev/vdb2 not large enough to join array (the disk#2 has been added to the machine afterwards)