LinuxQuestions.org
Visit Jeremy's Blog.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Server
User Name
Password
Linux - Server This forum is for the discussion of Linux Software used in a server related context.

Notices


Reply
  Search this Thread
Old 04-22-2015, 07:05 AM   #16
jpollard
Senior Member
 
Registered: Dec 2012
Location: Washington DC area
Distribution: Fedora, CentOS, Slackware
Posts: 4,912

Rep: Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513

An "inode mismatch" is trivial to fix. That is what fsck is for.

#3 is the only borderline --- and if you have a corrupted filesystem, then you usually have physical damage. Otherwise the usual fsck will repair it.

Now the rate of corruption depends on what filesystem you have on there. Ext3/4 are very good - I haven't seen a filesystem corruption in about 10 years of use with those. XFS is also very good. BTRFS (no experience),

I have used many disks without partitioning.. Usually in a NAS, and then partitioning the NAS created volume. But 16TB in a filesystem is not unworkable. Even 100TB and PB sizes are doable. None of those use partitioning as they are focused on the maximum amount of storage to be provided.

#1,2,4 are separate issues, and don't affect the risk of loosing data. Putting them on separate disks would.

Last edited by jpollard; 04-22-2015 at 07:10 AM.
 
Old 04-22-2015, 07:21 AM   #17
T3RM1NVT0R
Senior Member
 
Registered: Dec 2010
Location: Internet
Distribution: Linux Mint, SLES, CentOS, Red Hat
Posts: 2,385

Rep: Reputation: 477Reputation: 477Reputation: 477Reputation: 477Reputation: 477
I know inode mismatch is trivial that was just an example. The point here is if one partition get corruption I can separate it from other and only one set of data is affected. If a disk get corruption I am risking all data residing under that.

Another example I can give is about superblock corruption. We tried mounting it with backup superblock but it gave up. Finally we ran fsck to fix it but that didn't work. Red Hat then performed fsck (not sure the switches they used) but that ended up partition left with little to no data in it. Think of this happening with single disk scenario.
 
Old 04-22-2015, 07:59 AM   #18
jpollard
Senior Member
 
Registered: Dec 2012
Location: Washington DC area
Distribution: Fedora, CentOS, Slackware
Posts: 4,912

Rep: Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513
Same problem as when the partition table gets corrupted. You loose everything.

It is after all, the same blocks being used.

As I said, the risks are the same.

Last edited by jpollard; 04-22-2015 at 08:01 AM.
 
Old 04-22-2015, 08:09 AM   #19
T3RM1NVT0R
Senior Member
 
Registered: Dec 2010
Location: Internet
Distribution: Linux Mint, SLES, CentOS, Red Hat
Posts: 2,385

Rep: Reputation: 477Reputation: 477Reputation: 477Reputation: 477Reputation: 477
Partition table is different and superblocks for each partition are different. Superblocks are created when you format a partition. If partition table is corrupted all the partitions on that particular disk will get affected. However, if partition superblock is corrupted only that partition will get affected.
 
Old 04-22-2015, 08:11 AM   #20
adel87
Member
 
Registered: Nov 2014
Posts: 37

Original Poster
Rep: Reputation: Disabled
Smile

hi thank you.
 
Old 04-22-2015, 08:24 AM   #21
jpollard
Senior Member
 
Registered: Dec 2012
Location: Washington DC area
Distribution: Fedora, CentOS, Slackware
Posts: 4,912

Rep: Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513
Quote:
Originally Posted by T3RM1NVT0R View Post
Partition table is different and superblocks for each partition are different. Superblocks are created when you format a partition. If partition table is corrupted all the partitions on that particular disk will get affected. However, if partition superblock is corrupted only that partition will get affected.
The point is the block on disk is the same.

Without a partition table the home block is in a block that would have been used for a partition table.
 
Old 04-22-2015, 08:35 AM   #22
T3RM1NVT0R
Senior Member
 
Registered: Dec 2010
Location: Internet
Distribution: Linux Mint, SLES, CentOS, Red Hat
Posts: 2,385

Rep: Reputation: 477Reputation: 477Reputation: 477Reputation: 477Reputation: 477
Quote:
Originally Posted by jpollard View Post
The point is the block on disk is the same.

Without a partition table the home block is in a block that would have been used for a partition table.
Yes that is correct that is one scenario wherein partition table itself is corrupt. I think you are mixing two things. What I am trying to say is in case of a single disk a superblock corruption will bring down the whole disk. Whereas in case of partitioned disk superblock corruption will bring down only the partition for which it belongs to.

Simple words, partition table corruption will affect the whole disk. Superblock corruption will not affect the whole disk. Those are two different scenarios.
 
Old 04-22-2015, 08:40 AM   #23
jpollard
Senior Member
 
Registered: Dec 2012
Location: Washington DC area
Distribution: Fedora, CentOS, Slackware
Posts: 4,912

Rep: Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513Reputation: 1513
And both can affect the entire disk.

Partitioning is only for administrative use. It does not reduce the risk of a disk failure.
 
Old 04-22-2015, 09:50 AM   #24
T3RM1NVT0R
Senior Member
 
Registered: Dec 2010
Location: Internet
Distribution: Linux Mint, SLES, CentOS, Red Hat
Posts: 2,385

Rep: Reputation: 477Reputation: 477Reputation: 477Reputation: 477Reputation: 477
Quote:
Originally Posted by jpollard View Post
And both can affect the entire disk.

Partitioning is only for administrative use. It does not reduce the risk of a disk failure.
Superblock can affect entire disk incase you are using the whole disk as a single disk (without partitioning). Incase you are using a disk with partitions each partition will have it's own superblock and incase of superblock corruption only that partition will get affected.

Incase of hardware damage or partition table corruption entire disk will get affected irrespective of you are using disk with partitions or as single disk.

Here is the output from my test machine and that might help:

Code:
[root@rhel6-test ~]# dumpe2fs /dev/vda1 | grep superblock
dumpe2fs 1.41.12 (17-May-2010)
  Primary superblock at 1, Group descriptors at 2-3
  Backup superblock at 8193, Group descriptors at 8194-8195
  Backup superblock at 24577, Group descriptors at 24578-24579
  Backup superblock at 40961, Group descriptors at 40962-40963
  Backup superblock at 57345, Group descriptors at 57346-57347
  Backup superblock at 73729, Group descriptors at 73730-73731
  Backup superblock at 204801, Group descriptors at 204802-204803
  Backup superblock at 221185, Group descriptors at 221186-221187
  Backup superblock at 401409, Group descriptors at 401410-401411
[root@rhel6-test ~]# dumpe2fs /dev/mapper/vg_rhel6test-lv_root | grep superblock
dumpe2fs 1.41.12 (17-May-2010)
  Primary superblock at 0, Group descriptors at 1-1
  Backup superblock at 32768, Group descriptors at 32769-32769
  Backup superblock at 98304, Group descriptors at 98305-98305
  Backup superblock at 163840, Group descriptors at 163841-163841
  Backup superblock at 229376, Group descriptors at 229377-229377
  Backup superblock at 294912, Group descriptors at 294913-294913
  Backup superblock at 819200, Group descriptors at 819201-819201
  Backup superblock at 884736, Group descriptors at 884737-884737
  Backup superblock at 1605632, Group descriptors at 1605633-1605633
  Backup superblock at 2654208, Group descriptors at 2654209-2654209
[root@rhel6-test ~]# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_rhel6test-lv_root
                       12G  5.7G  5.2G  53% /
tmpfs                 751M   72K  751M   1% /dev/shm
/dev/vda1             485M   33M  427M   8% /boot
[root@rhel6-test ~]# mount /dev/vdb /data
[root@rhel6-test ~]# df -h
Filesystem            Size  Used Avail Use% Mounted on
/dev/mapper/vg_rhel6test-lv_root
                       12G  5.7G  5.2G  53% /
tmpfs                 751M   72K  751M   1% /dev/shm
/dev/vda1             485M   33M  427M   8% /boot
/dev/vdb              5.0G  138M  4.6G   3% /data
[root@rhel6-test ~]# dumpe2fs /dev/vdb | grep superblock
dumpe2fs 1.41.12 (17-May-2010)
  Primary superblock at 0, Group descriptors at 1-1
  Backup superblock at 32768, Group descriptors at 32769-32769
  Backup superblock at 98304, Group descriptors at 98305-98305
  Backup superblock at 163840, Group descriptors at 163841-163841
  Backup superblock at 229376, Group descriptors at 229377-229377
  Backup superblock at 294912, Group descriptors at 294913-294913
  Backup superblock at 819200, Group descriptors at 819201-819201
  Backup superblock at 884736, Group descriptors at 884737-884737
[root@rhel6-test ~]#
Here I have two disks on the system one is /dev/vda and another is /dev/vdb. The disk /dev/vda has got two partitions /dev/vda1 and /dev/vda2, /dev/vda1 is /boot and /dev/vda2 is a Linux LVM partition. The second disk I am using as single disk and formatted it directly without creating any partitions. Here is the output from fdisk -l:

Code:
[root@rhel6-test ~]# fdisk -l

Disk /dev/vda: 16.1 GB, 16106127360 bytes
16 heads, 63 sectors/track, 31207 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00088b50

   Device Boot      Start         End      Blocks   Id  System
/dev/vda1   *           3        1018      512000   83  Linux
Partition 1 does not end on cylinder boundary.
/dev/vda2            1018       31208    15215616   8e  Linux LVM
Partition 2 does not end on cylinder boundary.

Disk /dev/vdb: 5368 MB, 5368709120 bytes
16 heads, 63 sectors/track, 10402 cylinders
Units = cylinders of 1008 * 512 = 516096 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000000


Disk /dev/mapper/vg_rhel6test-lv_root: 12.4 GB, 12423528448 bytes
255 heads, 63 sectors/track, 1510 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000000


Disk /dev/mapper/vg_rhel6test-lv_swap: 3154 MB, 3154116608 bytes
255 heads, 63 sectors/track, 383 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Sector size (logical/physical): 512 bytes / 512 bytes
I/O size (minimum/optimal): 512 bytes / 512 bytes
Disk identifier: 0x00000000

[root@rhel6-test ~]#
If you see the output from dumpe2fs it clearly defines different superblock for the partitions or lvs. Incase my /dev/vda2 has got superblock corruption it will affect my root_lv not my /boot. However, that is not the case with /dev/vdb. As I am using /dev/vdb as single disk it has got single set of superblocks, not different like incase of partitioned disk. I hope this helps.
 
Old 04-27-2015, 06:58 AM   #25
voleg
Member
 
Registered: Oct 2013
Distribution: RedHat CentOS Fedora SuSE
Posts: 354

Rep: Reputation: 51
If you talking about GFS2 over cluster of VM nodes, I'd recommend using DRBD.
Shared VMDK is not working so good as DRBD.
See my try here Building active-active RedHat 6 Cluster with GFS2 over DRBD
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
Cluster question takayama Linux - Server 4 06-21-2011 07:34 PM
[SOLVED] Cluster question bino25 Linux - Software 2 02-04-2011 07:49 AM
HA cluster question illNino Linux - Software 3 04-17-2009 04:32 PM
Cluster question: Is it possible to make a bulletproof cluster? ValidiusMaximus Linux - Software 1 09-06-2005 02:07 PM
RedHat cluster question trigger638 Linux - Enterprise 0 06-01-2005 06:29 PM

LinuxQuestions.org > Forums > Linux Forums > Linux - Server

All times are GMT -5. The time now is 08:40 PM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Open Source Consulting | Domain Registration