Enable HTT with dual xeon fileserver ?

Enable HTT with dual xeon fileserver ?

[raid]

I'm building a Dual Xeon Fileserver
i can enable HTT in the bios (so i have 4 logical processors)

what is the advise for a fileserver ?
(both running hardware raid5 (3ware) and software raid5 (on promise
controllers)

Re: Enable HTT with dual xeon fileserver ?

[Martin Eriksson]

----- Original Message -----
From: <raid@ddx.a2000.nu>
To: <linux-raid@vger.kernel.org>
Sent: Monday, September 16, 2002 11:02 PM
Subject: Enable HTT with dual xeon fileserver ?


> I'm building a Dual Xeon Fileserver
> i can enable HTT in the bios (so i have 4 logical processors)

I'm very sceptical that a fileserver would gain any performance from HT.

In fact, I don't really realize why you even need two CPU's? If you are
doing database lookups it's another thing though, also HT could give a
performance gain in that case (if it's supported by the linux kernel?).

> what is the advise for a fileserver ?
> (both running hardware raid5 (3ware) and software raid5 (on promise
> controllers)

A follow-up question: Does sw-raid5 checksumming use multiple CPUs in any
way?

/Martin

Re: Enable HTT with dual xeon fileserver ?

[Neil Brown]

On Wednesday September 18, nitrax@giron.wox.org wrote:
> 
> A follow-up question: Does sw-raid5 checksumming use multiple CPUs in any
> way?

It could if separate threads were submitting requests on the same
array, but I don't think it would be very likely.  Lots of the
checksumming is done by raid5d which is single threaded.

Ofcourse if you had 2 raid5 arrays, they could quite possibly use one
cpu each.

NeilBrown

Could not have the right MD device name on /proc/partitions

[bmoon]

Hello,

I am having very strange and serious problem, which I could not
have the right MD device name from /proc/partitions.

I created md1,2,3,4 but it shows "md0" for all as you can see
from the following "cat /proc/partitions".

arm-raid# mdadm -C /dev/md1 -l1 -n2 /dev/hdb5 /dev/hdd5
mdadm: array /dev/md1 started.
arm-raid# mdadm -C /dev/md2 -l1 -n2 /dev/hdb6 /dev/hdd6
mdadm: array /dev/md2 started.
arm-raid#mdadm -C /dev/md3 -l1 -n2 /dev/hda5 /dev/hdc5
mdadm: array /dev/md3 started.
arm-raid#mdadm -C /dev/md4 -l1 -n2 /dev/hda6 /dev/hdc6
mdadm: array /dev/md4 started.

arm-raid#cat /proc/mdstat
Personalities : [linear] [raid0] [raid1] [raid5] 
read_ahead 1024 sectors
md4 : active raid1 hdc6[1] hda6[0]
      1024000 blocks [2/2] [UU]
      
md3 : active raid1 hdc5[1] hda5[0]
      1024000 blocks [2/2] [UU]
      
md2 : active raid1 hdd6[1] hdb6[0]
      1024000 blocks [2/2] [UU]
      
md1 : active raid1 hdd5[1] hdb5[0]
      1024000 blocks [2/2] [UU]
      
unused devices: <none>

arm-raid#cat /proc/partitions
major minor  #blocks  name

   9     1    1024000 md0
   9     2    1024000 md0
   9     3    1024000 md0
   9     4    1024000 md0
   3     0  120060864 hda
   3     1      16600 hda1

Any clues? suggestions?

Thanks in advance,

Bo

Re: Could not have the right MD device name on /proc/partitions

[Neil Brown]

On Thursday September 19, bo@sosnetwork.net wrote:
> Hello,
> 
> I am having very strange and serious problem, which I could not
> have the right MD device name from /proc/partitions.

You didn't say which kernel you were running.  2.4? 2.5? i386? sparc?
arm?

NeilBrown

System halt when re-inserting a HotSwap SCSI in Soft RAID1

[Jens Arnfelt]

Hi There!

I have two scsi disk with idendical partitiontables (see below) 
installed on a SCA HotSwap enable Fujitsu-Siemens server.

As a test I've removed one of the disks and the /dev/md0 went in 
degraded mode as expected. (se snip from /var/log/messages below)

... but what about /dev/md1 which also have a partition on /dev/sdb 
which I've removed?????
Nothing. And when I check "/proc/mdstat" the /dev/md1 seames to be 
running without error.... strange.

Note:
---------
Later investigation have show, that /dev/md1 would go into degraded mode 
if I've coppyed some data to it. (See fstab below).
Problem is that I have intended /dev/md1 as a partition for swap.
This is nesseray to really have a High Avalible (HA) system.
---------

The real problems start when I re-insert the /dev/sda.
The systems stops all activity on HD after some errore in the 
/var/log/messages and a hard reset is only option.

The system can however boot up. A look at /proc/mdstat shows that the 
/dev/md0 (still) and /dev/md1 is now correctly set as working in 
degraded mode.
If I now run "raidhotadd /dev/md0 /dev/sdb1" and "raidhotadd /dev/md0 
/dev/sdb1" evething rebuilts nicesly.

Later experiments have shown that I have to power off the system, insert 
the disk and boot.
This is however not the intended use of a HotSwap system.


BTW
SCSI card: 53c1010 Ultra3 SCSI Adapter from Symbios
uname -a => "Linux ABI1 2.4.18-4GB #1 Wed Sep 18 16:33:24 CEST 2002 i686 
SuSE 8.0 Raid" with RAID1, jbd and ext3 compiled into the kernel.
RAID: raidtools-0.90-349




--------------- snip /var/log/messages -------
Sep 19 14:25:57 ABI1 kernel: SCSI disk error : host 0 channel 0 id 1 lun 
0 return code = 100ff
Sep 19 14:25:57 ABI1 kernel: I/O error: dev 08:11, sector 37992
Sep 19 14:25:57 ABI1 kernel: raid1: Disk failure on sdb1, disabling device.
Sep 19 14:25:57 ABI1 kernel: Operation continuing on 1 devices
Sep 19 14:25:57 ABI1 kernel: md: recovery thread got woken up ...
Sep 19 14:25:57 ABI1 kernel: md: updating md0 RAID superblock on device
Sep 19 14:25:57 ABI1 kernel: md: (skipping faulty sdb1 )
Sep 19 14:25:57 ABI1 kernel: md: sda1 [events: 00000036]<6>(write) 
sda1's sb offset: 6816640
Sep 19 14:25:57 ABI1 kernel: SCSI disk error : host 0 channel 0 id 1 lun 
0 return code = 100ff
Sep 19 14:25:57 ABI1 kernel: I/O error: dev 08:11, sector 38000
Sep 19 14:25:58 ABI1 kernel: md0: no spare disk to reconstruct array! -- 
continuing in degraded mode
Sep 19 14:25:58 ABI1 kernel: md: recovery thread finished ...
--------------- /var/log/messages snip -------




----------------- output of "cat /proc/mdstat" before failer start 
--------------------
Personalities : [raid0] [raid1] [raid5]
read_ahead 1024 sectors
md0 : active raid1 sdb1[1] sda1[0]
6816640 blocks [2/2] [UU]

md1 : active raid1 sdb2[1] sda2[0]
2107328 blocks [2/2] [UU]
----------------- output of "cat /proc/mdstat" before failer end 
--------------------

----------------- output of "cat /proc/mdstat" AFTER failer 
start--------------------
Personalities : [raid0] [raid1] [raid5]
read_ahead 1024 sectors
md0 : active raid1 sdb1[1](F) sda1[0]
6816640 blocks [2/1] [U_]

md1 : active raid1 sdb2[1] sda2[0]
2107328 blocks [2/2] [UU]

unused devices: <none>
----------------- output of "cat /proc/mdstat" before AFTER end 
--------------------




----------------- output of "sfdisk -cl" start --------------------
Disk /dev/sda: 8715 cylinders, 64 heads, 32 sectors/track
Units = cylinders of 1048576 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System
/dev/sda1 0+ 6656 6657- 6816752 fd Linux raid autodetect
/dev/sda2 6657 8714 2058 2107392 fd Linux raid autodetect
/dev/sda3 0 - 0 0 0 Empty
/dev/sda4 0 - 0 0 0 Empty

Disk /dev/sdb: 8715 cylinders, 64 heads, 32 sectors/track
Units = cylinders of 1048576 bytes, blocks of 1024 bytes, counting from 0

Device Boot Start End #cyls #blocks Id System
/dev/sdb1 0+ 6656 6657- 6816752 fd Linux raid autodetect
/dev/sdb2 6657 8714 2058 2107392 fd Linux raid autodetect
/dev/sdb3 0 - 0 0 0 Empty
/dev/sdb4 0 - 0 0 0 Empty
----------------- output of "sfdisk -cl" end--------------------



------ raidtab start ------------
raiddev /dev/md0
raid-level 1
nr-raid-disks 2
nr-spare-disks 0
persistent-superblock 1
chunk-size 4
device /dev/sda1
raid-disk 0
device /dev/sdb1
raid-disk 1

raiddev /dev/md1
raid-level 1
nr-raid-disks 2
nr-spare-disks 0
persistent-superblock 1
chunk-size 4
device /dev/sda2
raid-disk 0
device /dev/sdb2
raid-disk 1
------ raidtab end ------------


----------- /etc/fstab start -----------
/dev/md0 / ext3 defaults 1 2
/dev/md1 /data ext3 defaults 1 2
devpts /dev/pts devpts defaults 0 0
/dev/cdrom /media/cdrom auto ro,noauto,user,exec 0 0
/dev/dvd /media/dvd auto ro,noauto,user,exec 0 0
/dev/fd0 /media/floppy auto noauto,user,sync 0 0
usbdevfs /proc/bus/usb usbdevfs noauto 0 0
proc /proc proc defaults 0 0
----------- /etc/fstab end -----------

PS.. Sorry for the spelling ;-D

Re: System halt when re-inserting a HotSwap SCSI in Soft RAID1

[Mads Peter Bach]

Jens Arnfelt wrote:

> Hi There!
>
> I have two scsi disk with idendical partitiontables (see below) 
> installed on a SCA HotSwap enable Fujitsu-Siemens server.
>
> As a test I've removed one of the disks and the /dev/md0 went in 
> degraded mode as expected. (se snip from /var/log/messages below)

[snip]

> The real problems start when I re-insert the /dev/sda.
> The systems stops all activity on HD after some errore in the 
> /var/log/messages and a hard reset is only option.


This isn't a soft raid problem, it's a SCSI layer issue:
(from the HOWTO)
"Hot-swapping with SCA

With SCA, it should be possible to hot-plug devices. However, I don't 
have the hardware to try this out, and I haven't heard from anyone who's 
tried, so I can't really give any recipe on how to do this.

If you want to play with this, you should know about SCSI and RAID 
internals anyway. So I'm not going to write something here that I can't 
verify works, instead I can give a few clues:

    * Grep for remove-single-device in linux/drivers/scsi/scsi.c
    * Take a look at raidhotremove and raidhotadd
"

Regards,
Mads Bach

RAID as backup ...

[C.Lee Taylor]

Greetings ...

	I intend to use RAID IDE to replace old, slow and problematic tape backup.

	This is just theory at the moment, and I would like you to help punch wholes 
in my idea ...

	Using normal IDE drives (3), USB 2.0 and an external housing and 5.25" 
mobile racks (2), I am playing with creating a RAID 1 setup.

	If I remove one of the IDE drives from my RAID system.  Could I put it way 
in a safe and use it as a backup?

Mailed
Lee

Re: RAID as backup ...

[Alvin Oga]

hi ya lee

- like the idea of using ide disks to backup other ide disks..

- i would NOT use raid1 ( mirroring ) for backup data
	- i would rather have 2 backup servers with 2 sets of backup disks
	than one backup server w/ mirror disks

- backup disks should be offline and unmountable except during the backup
 process

- if you use raid1, and take one disk and hide it, your backup is
  basically rendered dead(no longer raid) till you put that disk back in
  again to backup more data to the "raid1" backup system
	- just use the one disk in that same removable tray and
	you are more likely to have a suitable/usable backup system

- always have different backup disks for daily backup, weekly backups
  monthly backups and full backups... never use the same disks for it all
	( lose that disk, and you lose everything )

http://www.Linux-Backup.net
	- free backup scripts
	- 8x 120GB ide disks in one 1U chassis --> lots of fun(pain)

c ya
alvin

> 	I intend to use RAID IDE to replace old, slow and problematic tape backup.
> 
> 	This is just theory at the moment, and I would like you to help punch wholes 
> in my idea ...
> 
> 	Using normal IDE drives (3), USB 2.0 and an external housing and 5.25" 
> mobile racks (2), I am playing with creating a RAID 1 setup.
> 
> 	If I remove one of the IDE drives from my RAID system.  Could I put it way 
> in a safe and use it as a backup?
> 
> Mailed
> Lee
> 

Re: RAID as backup ...

[Franc Carter]

Only partially relevant, but . . .

We have just set up ide raid5 set which we rsync our critical
servers to each night, in a weekly rotation. As there is a bit
of distance between the rsync server and the raid5 machine we
effectively have a very cheap warm offsite backup.

It saved our bacon just recently.

On Sat, 21 Sep 2002 07:01, Alvin Oga wrote:
> hi ya lee
> 
> - like the idea of using ide disks to backup other ide disks..
> 
> - i would NOT use raid1 ( mirroring ) for backup data
> 	- i would rather have 2 backup servers with 2 sets of backup disks
> 	than one backup server w/ mirror disks
> 
> - backup disks should be offline and unmountable except during the backup
>  process
> 
> - if you use raid1, and take one disk and hide it, your backup is
>   basically rendered dead(no longer raid) till you put that disk back in
>   again to backup more data to the "raid1" backup system
> 	- just use the one disk in that same removable tray and
> 	you are more likely to have a suitable/usable backup system
> 
> - always have different backup disks for daily backup, weekly backups
>   monthly backups and full backups... never use the same disks for it all
> 	( lose that disk, and you lose everything )
> 
> http://www.Linux-Backup.net
> 	- free backup scripts
> 	- 8x 120GB ide disks in one 1U chassis --> lots of fun(pain)
> 
> c ya
> alvin
> 
> > 	I intend to use RAID IDE to replace old, slow and problematic tape 
backup.
> > 
> > 	This is just theory at the moment, and I would like you to help punch 
wholes 
> > in my idea ...
> > 
> > 	Using normal IDE drives (3), USB 2.0 and an external housing and 5.25" 
> > mobile racks (2), I am playing with creating a RAID 1 setup.
> > 
> > 	If I remove one of the IDE drives from my RAID system.  Could I put it 
way 
> > in a safe and use it as a backup?
> > 
> > Mailed
> > Lee
> > 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> 
> 

-- 
Franc Carter     Ph:61-2-8374-5071      Fax: 61-2-8374-5070
Systems Manager, SIRCA Ltd              http://www.sirca.org.au/

DISCLAIMER: The contents of this email, inclusive of attachments, may
be legally privileged and confidential. Any unauthorised use of the
contents is expressly prohibited. If you have received this message in
error or are not the intended recipient, you should destroy the email
message along with any attachment(s).  Unintended recipients of this
email are prohibited from retaining, disclosing, distributing or using
any information contained herein.  This email is also subject to
copyright. No part of it should be reproduced, adapted or transmitted
without the written consent of the copyright owner.

Re: RAID as backup ...

[C.Lee Taylor]

Alvin Oga wrote:
> hi ya lee
	Greetings ...

> - like the idea of using ide disks to backup other ide disks..
	Myself, IDE drives have become so cheap, that to purchase Tapes of the same 
size cost the same, plus if hardware ( tape drive ) is stolen ( whole 
server ) or destoried ( thinking of buring down the computer room ;-} ) you 
might have a problem replacing the Tape Drive and good tape drives also 
cost a fortune, which could easily put this inplace.  You backups then 
worth zero, and all the tapes big chunk of platic.

	If I use IDE drives, I can move these drives into workstations as bigger 
hard drives come out and put the newer bigger drives into my backup pool. 
Seems easy enough.

> - i would NOT use raid1 ( mirroring ) for backup data
> 	- i would rather have 2 backup servers with 2 sets of backup disks
> 	than one backup server w/ mirror disks
	Problem been, that to put down another server would cost large some of 
money, the idea behide RAID is inexpensive devices.

> - backup disks should be offline and unmountable except during the backup
>  process
> - if you use raid1, and take one disk and hide it, your backup is
>   basically rendered dead(no longer raid) till you put that disk back in
>   again to backup more data to the "raid1" backup system
> 	- just use the one disk in that same removable tray and
> 	you are more likely to have a suitable/usable backup system
	I throught about using hotplug scripts to backup the system when the drive 
is brought online, but this would be I would have to make sure that no 
files are open at the time we start the backup and then this could mean the 
server should not be used while this is happening, which means down time.

	Also, ( I am still testing so I am not 100% sure yet ), if running system 
dies, I could just put the last disk in as the primary drive and hope it 
boots ( as I said, not finished testing ), no install and restore.

> - always have different backup disks for daily backup, weekly backups
>   monthly backups and full backups... never use the same disks for it all
> 	( lose that disk, and you lose everything )
	Well, it up to the site on how many differant backups they want or can 
afford.  All that needs to be done, is the new disk need to be prepaired 
before use.

> http://www.Linux-Backup.net
> 	- free backup scripts
> 	- 8x 120GB ide disks in one 1U chassis --> lots of fun(pain)
	Will look at this site, thanks.


>>	I intend to use RAID IDE to replace old, slow and problematic tape backup.
>>
>>	This is just theory at the moment, and I would like you to help punch wholes 
>>in my idea ...
>>
>>	Using normal IDE drives (3), USB 2.0 and an external housing and 5.25" 
>>mobile racks (2), I am playing with creating a RAID 1 setup.
>>
>>	If I remove one of the IDE drives from my RAID system.  Could I put it way 
>>in a safe and use it as a backup?


Mailed
Lee

Re: Enable HTT with dual xeon fileserver ?

[raid]

On Wed, 18 Sep 2002, Martin Eriksson wrote:
> ----- Original Message -----
> From: <raid@ddx.a2000.nu>
> To: <linux-raid@vger.kernel.org>
> Sent: Monday, September 16, 2002 11:02 PM
> Subject: Enable HTT with dual xeon fileserver ?
>
> > I'm building a Dual Xeon Fileserver
> > i can enable HTT in the bios (so i have 4 logical processors)
>
> I'm very sceptical that a fileserver would gain any performance from HT.
don't know, just wondering
maybe samba/nfs can use htt/dual cpu ?

> In fact, I don't really realize why you even need two CPU's? If you are
i also don't think i really need 2 cpu's
but i bought the second one, because it was just a bit more (compared to
the whole price of the machine)


> doing database lookups it's another thing though, also HT could give a
> performance gain in that case (if it's supported by the linux kernel?).
it is supported

Re: Enable HTT with dual xeon fileserver ?

[Martin Eriksson]

----- Original Message -----
From: <raid@ddx.a2000.nu>
To: "Martin Eriksson" <nitrax@giron.wox.org>
Cc: <linux-raid@vger.kernel.org>
Sent: Wednesday, September 18, 2002 8:08 PM
Subject: Re: Enable HTT with dual xeon fileserver ?


> On Wed, 18 Sep 2002, Martin Eriksson wrote:
> > ----- Original Message -----
> > From: <raid@ddx.a2000.nu>
> > To: <linux-raid@vger.kernel.org>
> > Sent: Monday, September 16, 2002 11:02 PM
> > Subject: Enable HTT with dual xeon fileserver ?
> >
> > > I'm building a Dual Xeon Fileserver
> > > i can enable HTT in the bios (so i have 4 logical processors)
> >
> > I'm very sceptical that a fileserver would gain any performance from HT.
> don't know, just wondering
> maybe samba/nfs can use htt/dual cpu ?


Yes, theoretically. Your "problem" is still that all your disk arrays and
especially network, are slower than *one* of your CPUs. Well that isn't
really a problem, you propably have a kick*ss server in any case.

> > In fact, I don't really realize why you even need two CPU's? If you are
> i also don't think i really need 2 cpu's
> but i bought the second one, because it was just a bit more (compared to
> the whole price of the machine)

Ok, that's reasonable.

/Martin