Koozali.org: home of the SME Server

Help needed with this

Janm

Help needed with this
« on: April 13, 2006, 04:56:38 PM »
edited

Jan

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #1 on: April 15, 2006, 01:41:49 PM »
What is it you need help with?

Offline Tib

  • *
  • 571
  • +0/-0
    • http://www.tibors.net
Help needed with this
« Reply #2 on: April 15, 2006, 02:09:32 PM »
This looks more like an advert to me


Regards,

Tib

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #3 on: April 15, 2006, 03:30:11 PM »
true... and it is not the SME server...
Look at this old thread:
http://forums.contribs.org/index.php?topic=27314.0

Janm

Sorry
« Reply #4 on: April 15, 2006, 08:13:37 PM »
Sorry for the misleding picture on
my site now
This was the original post i would show

SME Server 7.0rc1   Copyright (C) 1999-2006 Mitel Corporation
Disk redundancy status as of Saturday April 15, 2006 20:05:32
Current RAID status:
Personalities : [raid1] [raid5]
md2 : active raid5 sdc2[2] sdb2[1] sda2[0]
70878208 blocks level 5, 256k chunk, algorithm 2 [3/3] [UUU]
md1 : active raid1 sdc1[2] sdb1[1] sda1[0]
104320 blocks [3/3] [UUU]
unused devices: <none>
One or more devices is not RAID1
Manual intervention may be required.
<Ok>

Jan  :hammer:

By the way i have 3 different servers i test around with
thats why somtimes there is a different site comming up   :hammer:

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #5 on: April 16, 2006, 07:08:52 AM »
Doesn't RAID1 require an equal number of disks? you have 3.

Janm

Help needed with this
« Reply #6 on: April 16, 2006, 04:46:59 PM »
Quote from: "greg"
Doesn't RAID1 require an equal number of disks? you have 3.


I have read this

For Highest performance, the controller must be able to perform two concurrent separate Reads per mirrored pair or two duplicate Writes per mirrored pair.

RAID Level 1 requires a minimum of 2 drives to implement

One Write or two Reads possible per mirrored pair

Twice the Read transaction rate of single disks, same Write transaction rate as single disks


100% redundancy of data means no rebuild is necessary in case of a disk failure, just a copy to the replacement disk


Transfer rate per block is equal to that of a single disk


Under certain circumstances, RAID 1 can sustain multiple simultaneous drive failures


Simplest RAID storage subsystem design

Highest disk overhead of all RAID types (100%) - inefficient

Typically the RAID function is done by system software, loading the CPU/Server and possibly degrading throughput at high activity levels. Hardware implementation is strongly recommended


May not support hot swap of failed disk when implemented in "software"

Ihave 3 disks on in hardware raid adaptec 7902

[root@www ~]# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   2588 MB in  2.00 seconds = 1294.20 MB/sec
 Timing buffered disk reads:  158 MB in  3.01 seconds =  52.50 MB/sec
[root@www ~]# hdparm -tT /dev/sdb

/dev/sdb:
 Timing cached reads:   2752 MB in  2.00 seconds = 1374.15 MB/sec
 Timing buffered disk reads:  160 MB in  3.00 seconds =  53.25 MB/sec
[root@www ~]# hdparm -tT /dev/sdc

/dev/sdc:
 Timing cached reads:   2608 MB in  2.00 seconds = 1304.20 MB/sec
 Timing buffered disk reads:  158 MB in  3.04 seconds =  52.05 MB/sec
[root@www ~]#



The real Problem is i am trying to get it working only in hardware
without any luck  :evil:

Regards Jan

Offline pfloor

  • *****
  • 889
  • +1/-0
Help needed with this
« Reply #7 on: April 16, 2006, 07:00:21 PM »
Quote from: "greg"
Doesn't RAID1 require an equal number of disks? you have 3.
No.

SME software raid is designed for any number of disks "of equal size".

2 disks will create a raid 1 pair.
3+ disks will create a raid 5
5+ disks will create a raid 6

However, the boot partition will always be a raid 1 no matter how many disks you have.  Linux can only boot off of no raid or SW raid 1.

Janm,
There is nothing wrong with your raid array.  It is reporting properly.  The reports in the console are being refined at this time and should be corrected soon.

There are at least 25 bugs regarding SW RAID in the bug tracker and probably 10 of them are about how it reports.  There is an ongoing effort to make the raid arrays more robust including but not limited to:

1-Better reporting.
2-Proper automatic rebuilding of replaced failed disks.
3-Proper conversion from SME6
4-Proper detection of existing SME6 raid pairs.

The dev team is dedicated to making upgrades from SME6 work and this RAID issue has been a challange.  SME7 added support for raid5/6 and it is VERY complicated.  Shad, Charlie and Gordon (and many others) have put in countless hours getting RAID1/5/6 to work.  It is close, but not perfect...YET!!!

Paul
In life, you must either "Push, Pull or Get out of the way!"

Offline pfloor

  • *****
  • 889
  • +1/-0
Help needed with this
« Reply #8 on: April 16, 2006, 07:15:44 PM »
Quote from: "Janm"
Ihave 3 disks on in hardware raid adaptec 7902

[root@www ~]# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   2588 MB in  2.00 seconds = 1294.20 MB/sec
 Timing buffered disk reads:  158 MB in  3.01 seconds =  52.50 MB/sec
[root@www ~]# hdparm -tT /dev/sdb

/dev/sdb:
 Timing cached reads:   2752 MB in  2.00 seconds = 1374.15 MB/sec
 Timing buffered disk reads:  160 MB in  3.00 seconds =  53.25 MB/sec
[root@www ~]# hdparm -tT /dev/sdc

/dev/sdc:
 Timing cached reads:   2608 MB in  2.00 seconds = 1304.20 MB/sec
 Timing buffered disk reads:  158 MB in  3.04 seconds =  52.05 MB/sec
[root@www ~]#


If you have a "hardware" raid then SME should only see 1 disk regardless of how many are attached to the controller. SME has detected the 3 disks and set them up properly as software raid5.  Either your controller card is not supported and/or is a "Fake Raid" card.

Quote from: "Janm"
take a look at the pic

http://design.jm-data.dk/

The real Problem is i am trying to get it working only in hardware
without any luck  :evil:

Regards Jan


This link still does not contain anything (especially a pic) about your problem.
In life, you must either "Push, Pull or Get out of the way!"

Janm

Thanks pflor
« Reply #9 on: April 16, 2006, 07:45:42 PM »
Thanks Pflor for your reply
the server i use is
Ibm :  eServ 226 Xseries 226 tower
Model :MT-M 8488-5AY
dual xeon 3.2 ghz 2 mb cache
with factory build adaptec scsi Aic 7902b raidbus controller
and scsi u320 interupt controller

Hope i will understand it better after your reply
i am reading all i can get for now thanks again
for all the patient

Jan

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #10 on: April 16, 2006, 08:02:11 PM »
pfloor wrote:
Quote
No.

SME software raid is designed for any number of disks "of equal size".

2 disks will create a raid 1 pair.
3+ disks will create a raid 5
5+ disks will create a raid 6


pfloor, true - but real raid1 requires an equal number of disks.


janm wrote:

Quote


dev/sda:
Timing cached reads: 2588 MB in 2.00 seconds = 1294.20 MB/sec
Timing buffered disk reads: 158 MB in 3.01 seconds = 52.50 MB/sec
[root@www ~]# hdparm -tT /dev/sdb

/dev/sdb:
Timing cached reads: 2752 MB in 2.00 seconds = 1374.15 MB/sec
Timing buffered disk reads: 160 MB in 3.00 seconds = 53.25 MB/sec
[root@www ~]# hdparm -tT /dev/sdc

/dev/sdc:
Timing cached reads: 2608 MB in 2.00 seconds = 1304.20 MB/sec
Timing buffered disk reads: 158 MB in 3.04 seconds = 52.05 MB/sec
[root@www ~]#



Here it looks to me that your adaptec 7902 has not been properly configured in the system.
If it should work as a true hardware raid - you should only see one disk.
Could you try to enlarge the picture on your page - it is difficult to see the details..

/G

Janm

Help needed with this
« Reply #11 on: April 16, 2006, 08:39:28 PM »
What picture do you want made bigger the console or phpsysinfo
Jan

Offline pfloor

  • *****
  • 889
  • +1/-0
Help needed with this
« Reply #12 on: April 16, 2006, 08:58:17 PM »
Quote
pfloor, true - but real raid1 requires an equal number of disks.
I don't quite understand what you mean by "equal number of disks"..."Equal" to what?  Do you mean "even number of disks" as in 2, 4, 6?

RAID1 can be set up on 2 or more disks of equal size.  If you use different size disks, your array will be the size of the smallest disk.  You can use 3, 4, 5, 11, 99 disks, it doesn't matter as they will all just mirror each other.  Adding disks only adds redundancy and not storage space.

Traditionally RAID1 is set up on a pair of drives but can be set up on any number of drives and using any of the disks above the initial 2 disks as either "active" disks or "spare" disks in any combination.
In life, you must either "Push, Pull or Get out of the way!"

Janm

Help needed with this
« Reply #13 on: April 16, 2006, 09:20:21 PM »
i just tried this command
login as: root
root@www's password:
Last login: Sun Apr 16 21:13:56 2006 from pc-00146.design.jm-data.dk
CentOS release 4.2 (Final) - SME Server 7.0rc1
[root@www ~]# cat /proc/mdstat
Personalities : [raid1] [raid5]
md2 : active raid5 sdc2[1] sda2[0]
      70878208 blocks level 5, 256k chunk, algorithm 2 [3/2] [UU_]

md1 : active raid1 sdc1[1] sda1[0]
      104320 blocks [3/2] [UU_]

unused devices: <none>
[root@www ~]#

jan

Offline pfloor

  • *****
  • 889
  • +1/-0
Help needed with this
« Reply #14 on: April 16, 2006, 09:44:41 PM »
You have lost sdb in your raid array.

Also, I don't think your raid adapter is supported.  At least it was not supported in SME6 and may still dot be supported.

Paul
In life, you must either "Push, Pull or Get out of the way!"

Janm

ihanks
« Reply #15 on: April 16, 2006, 10:03:07 PM »
Quote from: "pfloor"
You have lost sdb in your raid array.

Also, I don't think your raid adapter is supported.  At least it was not supported in SME6 and may still dot be supported.

Paul



Thanks i will do a complete reinstall now to se if thats help
so for a while i have no server and no mail  :hammer:
and i will try to get the support file *.img for the server from ibms homesite now so it can see the raid

http://www.acnc.com/04_01_00.html


Jan have a nice evening and not to many  :pint:
by for now Jan

Janm

i am back again
« Reply #16 on: April 17, 2006, 12:00:28 AM »
Hej i am back again  :hammer:
edited

Grin
regards Jan   :pint:

Last login: Sun Apr 16 23:30:31 2006
CentOS release 4.2 (Final) - SME Server 7.0rc1
[root@www ~]# cat /proc/mdstat
Personalities : [raid1] [raid5]
md2 : active raid5 sdc2[2] sdb2[1] sda2[0]
      70878208 blocks level 5, 256k chunk, algorithm 2 [3/3] [UUU]

md1 : active raid1 sdc1[2] sdb1[1] sda1[0]
      104320 blocks [3/3] [UUU]

unused devices: <none>
[root@www ~]#

Jan :hammer:

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #17 on: April 17, 2006, 12:40:58 AM »
pfloor wrote:
Quote


I don't quite understand what you mean by "equal number of disks"..."Equal" to what? Do you mean "even number of disks" as in 2, 4, 6?

RAID1 can be set up on 2 or more disks of equal size. If you use different size disks, your array will be the size of the smallest disk. You can use 3, 4, 5, 11, 99 disks, it doesn't matter as they will all just mirror each other. Adding disks only adds redundancy and not storage space.

Traditionally RAID1 is set up on a pair of drives but can be set up on any number of drives and using any of the disks above the initial 2 disks as either "active" disks or "spare" disks in any combination.


Sure - I mean even. And I also know that it's just mirroring.
RAID1 is usually made pairwise with the same size disks. It can also be striped to RAID1+0 or RAID10.

/G

Offline pfloor

  • *****
  • 889
  • +1/-0
Help needed with this
« Reply #18 on: April 17, 2006, 03:06:36 AM »
Quote from: "greg"
Sure - I mean even. And I also know that it's just mirroring.
RAID1 is usually made pairwise with the same size disks.


So then, is an *even* number of disks "usual" or "reqiured"?  You have changed your statement here.

The fact of the matter here is that you stated:

"pfloor, true - but real raid1 requires an equal number of disks. "

and this statement is just down right untrue.  Raid1 only has 1 real "requirement" and that would be that the raid1 array contains at least 2 partitions (not even 2 seperate disks).  You can mirror 2 (or more) partitions on the same disk if you really wanted to.  Logic tells us however that would not be a very good thing to do.

Mirroring just gives us a better chance of NOT losing our data. Lets say that 1 in 500 drives failed within 3 years.  That would mean that you have a 1 in 500 chance of losing your data.

1 disk = 1 in 500 chance of total data loss
2 disk mirror = 1 in 250,000 chance of total data loss
3 disk mirror = 1 in 125,000,000 chance of total data loss
4 disk mirror = 1 in 62,500,000,000 chance of total data loss
5 disk mirror = 1 in 31,250,000,000,000 chance of total data loss (that is a one 31 quadrillion chance)

Mirroring is the BEST data redundancy for mission critical situations.  In a 5 disk mirror, you would have to have all 5 disks fail all at once to lose your data...This is almost statistically impossible.

In contrast, a raid5 aray (regardless of the number of disks) can only withstand the loss of 1 disk.  This puts the raid5 array at a 1 in 250,000 chance of total data loss.  Redundancy versus performance...You choose.

Now back to the OP's problem...
In life, you must either "Push, Pull or Get out of the way!"

Offline pfloor

  • *****
  • 889
  • +1/-0
Re: i am back again
« Reply #19 on: April 17, 2006, 03:19:33 AM »
Quote from: "Janm"
[root@www ~]# cat /proc/mdstat
Personalities : [raid1] [raid5]
md2 : active raid5 sdc2[2] sdb2[1] sda2[0]
      70878208 blocks level 5, 256k chunk, algorithm 2 [3/3] [UUU]

md1 : active raid1 sdc1[2] sdb1[1] sda1[0]
      104320 blocks [3/3] [UUU]

unused devices: <none>
[root@www ~]#

Jan :hammer:

Jan, your hardware raid card is NOT being recognized by SME.  mdstat is the status of the software raid that SME set up during the installation.  It shows 3 drives and 2 partitions on each drive.

Partitions sda1, sdb1 and sdc1 are set up as a raid1 and running fine.  This is where the boot partition is kept so that SME can boot.

Partitions sda2, sdb2 and sdc2 are the rest of the drives set up as a raid5.  This is also running properly.  This is where the rest of the files are being stored.

This is actually 2 seperate raid devices (one raid1 and one raid5) running on the same 3 drives, just different partitions.

SME has taken over your raid controller and is just using it as a plain old disk controller and performing software raid.

If the raid controller was working as a true raid controller, it would just present itself to SME as a single disk and SME would treat it as 1 disk and not 3.

Either you don't have the controller set up correctly or it is what is known as "Fake Raid".  You need to do some research and find out for yourself.

Paul
In life, you must either "Push, Pull or Get out of the way!"

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #20 on: April 17, 2006, 08:16:28 AM »
pfloor wrote:
Quote

....
1 disk = 1 in 500 chance of total data loss
2 disk mirror = 1 in 250,000 chance of total data loss
3 disk mirror = 1 in 125,000,000 chance of total data loss
4 disk mirror = 1 in 62,500,000,000 chance of total data loss
5 disk mirror = 1 in 31,250,000,000,000 chance of total data loss (that is a one 31 quadrillion chance)
....


OK, you got me on this. I surrender. I was not precise in my statements, and will be silent for now on this....

Also thanks for the lesson. I appreciate that.
However,  unless you have very mission critical data, more than two drives must be unusual - or what? I've never seen it in my career.

Should I take your point of view, that you (when it's just two mirroring RAID1 drives, or a couple using RAID1+0) would prefer RAID5?
In your small calculation above it gives the same possibility of fail as a pair RAID1 consisting of two drives - or, say, 6 drives in RAID1+0 config.

Cheers

/G
[/quote]

Offline greg

  • ***
  • 51
  • +0/-0
Help needed with this
« Reply #21 on: April 17, 2006, 09:02:51 AM »
Jan,

You can try to find a driver here
http://www-306.ibm.com/pc/support/site.wss/document.do?sitestyle=ibm&lndocid=MIGR-55285&velxr-layout=print

However, I'm not sure they work, as the latest OS I see is RHEL 3, and it may be for SATA devices...
Moreover, I'm not sure how to create the driver disks from these files and subsequently use the 'linux dd' option when setting up the system.

Can this actually be done with SME?

Also, if you search the web - you will see several people having your problem.
So - the easiest solution may be to buy a LSI MegaRaid or similar or let SME do the software RAID ...

/G

Offline pfloor

  • *****
  • 889
  • +1/-0
Help needed with this
« Reply #22 on: April 17, 2006, 06:10:48 PM »
Quote from: "greg"
Also thanks for the lesson. I appreciate that.
However,  unless you have very mission critical data, more than two drives must be unusual - or what? I've never seen it in my career.

More than 2 drives in a raid1 is VERY unusual but I have seen it in accounting firms.  They are *very* particular about data loss and don't care as much about speed.

Quote
Should I take your point of view, that you (when it's just two mirroring RAID1 drives, or a couple using RAID1+0) would prefer RAID5?

I really don't have a preference.  I ran 2 disk raid1 for a long time and felt very secure with it.  I just liked the idea of having "everything" doubled.  I currently run SME7 with a 4 disk raid5.  To be honest, it makes me a bit nervous, probably because I don't know enough about it and I really don't understand how it works...But that in itself is a whole new topic for discussion.

As for the OP's problem.  I looked around at issues revolving around his controller and I am convinced that it is not a real raid card.  Drivers, re-compiling kernels, etc... That card should be invisable to the whole machine and just act like a single drive.  To make it work is going to take cpu resources and he might as well just use the SW raid built into SME7.  SW raid is very good and actually out-performs most "fake raid" cards anyway.

Paul
In life, you must either "Push, Pull or Get out of the way!"

Janm

Help needed with this
« Reply #23 on: April 17, 2006, 10:44:55 PM »
Thanks again both of you
for your interest in my problems
i did a new install again today with only 2 drive in computer
Thanks again everybody

edited this post has removed the pictures now

all the best from Jan   :-D