[ictdude]

After a lot of testing and trial and error. OVH could not found a defect. I ordered a second new i7-2T. And also ordered a hardware raid. Gues what ? All performance problems are gone. Big change this i7-2T had some defects. The better give me a refund.. Am happy its done ppfff

[ictdude]

What does the BSOD screen give you, in terms of error messages? Windows is quite "sensitive" when it comes to hardware configuration (ie., the network driver selection in Proxmox, ACPI, etc).

I know what you mean. But i did a install like i also did on my other Proxbox. And this server gave problems other did not. I also did try to restore a VM from other server to this one. Normally this should work. But VM system hangs. Never had those kind of problems. Also a fresh Windows VM install on this server took way to long. If it works on other servers and i do the same steps it don't make sense that here i run in problems, don't you think ?

[Myatu]

What does the BSOD screen give you, in terms of error messages? Windows is quite "sensitive" when it comes to hardware configuration (ie., the network driver selection in Proxmox, ACPI, etc).

[ictdude]

I get about 22% max with one windows 2003 vm on the IODelay based in the webpanel.

Performance seems to be fine, along with response times. - Could possibly be a broken installation of faulty HD. I would put forward to support and see what they say.

My server performed much better after upgrade to Proxmox 1.4. But after the upgrade other problems. Like a BSOD when installing windows.
I gues i have to return the server and take a other one. They already exchanged the power unit because server was dead.. ppff .. I would like to take a other i7-2T but i dont like to waste money if i run again in those troubles. I know the Q-1T performance works fine but no raid a delemma.
Or find for a solution in the Proxmox forum. But if the mainboard dont work well maybe defects in this case then it will not help ..

Paw-Fox what hardware has your server ?

[ovhfan2]

Thanks, Myatu

[Myatu]

Does the functioning disk simply take over without you having to type in any commands?

There's no need to type any commands. Once a disk fails, the RAID becomes "degraded" like with a hardware raid. All the work is then done on the remaining HD.

If configured properly (see /etc/mdadm/mdadm.conf) you'll also receive an e-mail as soon as this happens, so you can know about this fairly quickly.

How long does it take for OVH to replace the failed disk?

In my case it was only one sector that went bad (and so caused inconsistencies between the two HDs). Sectors do go bad after a while, which is why HDs have a number of spares available. This is something I resolved myself, but OVH will not take long once they've confirmed a HD needs replacement.

Do you see a performance hit with software RAID?

It does affect the overall performance, but not by a whole lot. In my opinion, this slight loss of performance is made up by the extra security against loss of data. You can also reduce this performance loss by implementing memory (RAM) caching where possible.

Sorry for all the questions. I am considering using software RAID myself with one of the OVH's i920 offerings, but just not too sure if I'd need any technical knowhow to manage it...

It's essentially a "set and forget" feature. The only time where you need to get involved is when the RAID becomes degraded -- ie., what commands do you type to find out which HD failed, why did it fail and how can you fix it. But that's where this forum and Google come in handy

PS.: This is what an e-mail looks like:

This is an automatically generated mail message from mdadm
running on ...

A Fail event had been detected on md device /dev/md1.

It could be related to component device /dev/sdb1.

Faithfully yours, etc.

P.S. The /proc/mdstat file currently contains the following:

Personalities : [raid1] [raid0] [raid6] [raid5] [raid4] [linear] [multipath] [raid10] 
md1 : active raid1 sda1[0] sdb1[2](F)
      20482752 blocks [2/1] [U_]
      [===>.................]  check = 16.0% (3280064/20482752) finish=27.3min speed=10455K/sec
      
md2 : active raid1 sda2[0] sdb2[1]
      707896128 blocks [2/2] [UU]
      	resync=DELAYED
      
unused devices: 

It even tells you which HD failed (in this case, the problem was on the 2nd HD, marked by an (F), confirmed by smartctl).

[ovhfan2]

No, software.

Does the functioning disk simply take over without you having to type in any commands?

How long does it take for OVH to replace the failed disk?

Do you see a performance hit with software RAID?

Sorry for all the questions. I am considering using software RAID myself with one of the OVH's i920 offerings, but just not too sure if I'd need any technical knowhow to manage it...

[Myatu]

Was that hardware RAID?

No, software.

[ovhfan2]

............

(Mine's RAID 1 by default actually, and today is the reason why: HD failure But still running fine )

Was that hardware RAID?

[ictdude]

Maybe it will be better to try a upgrade to Proxmox VE 1.4
I see a lot of improvements looking to storage: http://pve.proxmox.com/wiki/Roadmap

http://pve.proxmox.com/wiki/Downloads

Also Updated kernel and some driver support.

[ictdude]

That is funny. After reading this I checked the status of my RAID 1, and it was showing as DEGRADED.
Removed the Disk showing as DEGRADED, added it back and are now rebuilding the RAID 1 on that Disk.
Hardware Raid FTW!!!

You have hardware raid on a 2T? What commands did you use it to check this ?

[ictdude]

RAID and the CPU is the only major difference between the two servers. Given the chance that it's the CPU is slim, I'd be suspicious about the RAID (drivers). Could be wrong though, because pveperf really doesn't reveal much useful stuff.

Did you go for the hardware raid option? (Mine's RAID 1 by default actually, and today is the reason why: HD failure But still running fine )

No software.. First like to test the server befor i go for more options..

Only professional options have hardware raid. Thats how i did understand it. Default no hardware raid controller. Spec say Software controlled ? Well some stuff is pulling down the performance.. The Q-1t dont have a big delay... See what the support will do about it. If they dont fix it i go back to a second Q-1T But 4 gigs ram Less and no raid Paw-Fox in this forum tells us he get max about 22% IO Delay. With the Q-1T this looks until now better.

You can test it by copy data. Q-1T give max 12 while i7-2T jumps to 60 % and then am not in a VM but from the Proxmox host. And all goes slow ...But thats what my server is showing.. drivers could be or hardware not funny..

I can see the mainboard is a product: X8STi vendor: Supermicro. This board has a onboard Sata intel ICH10R Controller Raid aha ?
How to tweek this Controller ? http://www.supermicro.com/products/m.../X58/X8STi.cfm

I could test it with out the software raid: Because OS Compatibility:
http://www.supermicro.com/support/re...s/OS/C7_X8.cfm --> Ubuntu 8 --> 4. SATA (w/o RAID, IDE enhance mode) Debian was not on the list so Ubuntu 8 will do..

*-ide:0
description: IDE interface
product: 82801JI (ICH10 Family) 4 port SATA IDE Controller
vendor: Intel Corporation
physical id: 1f.2
bus info: pci@0000:00:1f.2
logical name: scsi0
logical name: scsi1
version: 00
width: 32 bits
clock: 66MHz
capabilities: ide pm bus_master cap_list emulated
configuration: driver=ata_piix latency=0 module=ata_piix
*-disk:0
description: ATA Disk
product: WDC WD10EADS-00P
vendor: Western Digital
physical id: 0
bus info: scsi@0:0.0.0
logical name: /dev/sda
version: 01.0
serial: WD-WMAVU0331176
size: 931GiB (1TB)
capabilities: partitioned partitioned:dos
configuration: ansiversion=5 signature=0006dae3
*-volume:0
description: EXT3 volume
vendor: Linux
physical id: 1
bus info: scsi@0:0.0.0,1
logical name: /dev/sda1
version: 1.0
serial: 0e063eb1-92ba-4966-90ea-c8acdd1d9b95
size: 19GiB
capacity: 19GiB

---------------------------
My setup RAID 1 (Software)
---------------------------

root@ks:~# cat /proc/mdstat

Personalities : [raid1] [raid0] [raid6] [raid5] [raid4] [linear] [multipath] [raid10]

md2 : active raid1 sda2[0] sdb2[1]
955231168 blocks [2/2] [UU]

md1 : active raid1 sda1[0] sdb1[1]
20479936 blocks [2/2] [UU]

unused devices:

[derchris]

RAID and the CPU is the only major difference between the two servers. Given the chance that it's the CPU is slim, I'd be suspicious about the RAID (drivers). Could be wrong though, because pveperf really doesn't reveal much useful stuff.

Did you go for the hardware raid option? (Mine's RAID 1 by default actually, and today is the reason why: HD failure But still running fine )

That is funny. After reading this I checked the status of my RAID 1, and it was showing as DEGRADED.
Removed the Disk showing as DEGRADED, added it back and are now rebuilding the RAID 1 on that Disk.
Hardware Raid FTW!!!

[Myatu]

Its default OVH. Raid 0 striped. What status do you mean ?

RAID and the CPU is the only major difference between the two servers. Given the chance that it's the CPU is slim, I'd be suspicious about the RAID (drivers). Could be wrong though, because pveperf really doesn't reveal much useful stuff.

Did you go for the hardware raid option? (Mine's RAID 1 by default actually, and today is the reason why: HD failure But still running fine )

[Paw-Fox]

I get about 22% max with one windows 2003 vm on the IODelay based in the webpanel.

Performance seems to be fine, along with response times. - Could possibly be a broken installation of faulty HD. I would put forward to support and see what they say.

[ictdude]

Just got mine setup - Yay.

This is the stats I got from running pvepref - I haven't installed a VM yet, but will edit this post when I do.

root@ks366753:~# pveperf
CPU BOGOMIPS: 42671.21
REGEX/SECOND: 497196
HD SIZE: 4.96 GB (/dev/md1)
BUFFERED READS: 83.87 MB/sec
AVERAGE SEEK TIME: 10.05 ms
FSYNCS/SECOND: 988.88
DNS EXT: 21.39 ms
DNS INT: 17.44 ms (ovh.net)

How does the IO Delay looks like in the webgui ? And performance ?

[ictdude]

What's the status of your RAID?

Its default OVH. Raid 0 striped. What status do you mean ?

[Paw-Fox]

Just got mine setup - Yay.

This is the stats I got from running pvepref - I haven't installed a VM yet, but will edit this post when I do.

root@ks366753:~# pveperf
CPU BOGOMIPS: 42671.21
REGEX/SECOND: 497196
HD SIZE: 4.96 GB (/dev/md1)
BUFFERED READS: 83.87 MB/sec
AVERAGE SEEK TIME: 10.05 ms
FSYNCS/SECOND: 988.88
DNS EXT: 21.39 ms
DNS INT: 17.44 ms (ovh.net)

[Myatu]

What's the status of your RAID?

[ictdude]

Hope this isn't true, as I just ordered the 2T - but I will be happy to test once its up.

I did send this to the OVH support desk. See what they will do about it.
Its not funny to get a 2T while the 1T gets they job done and the 2T not. Could be hardware related, We will see. Let me know if you have youre 2T
running and what the performance will be .

[Paw-Fox]

Hope this isn't true, as I just ordered the 2T - but I will be happy to test once its up.

[ictdude]

Bad performance Proxmox on i7-2T while Q-1T is great ?

I have a i7-2T kimsufi and use it for Proxmox. This server response slow and have a lot of times high IO Delay with Proxmox.And only one VM installed and almost is a sleep. When using the webinterface Proxmox a lot of times slow response. Did some hardware tests in rescue pro mode. And test say all is ok ?

I also have a Q-1T kimsufi with more VM's and this one have a real good performance. Also when VM's are really active. This is really strange. Is the i7-2T not as good (reallty bad) as a Q-1T or could there be hardware failures ? I already did reinstall the system result is the same ?
Then i better get a second Q-1T and forget about the i7-2T but wasted some money here...

Q-1T kimsufi

root@fr:~# pveperf
CPU BOGOMIPS: 19239.08
REGEX/SECOND: 770642
HD SIZE: 4.92 GB (/dev/sda1)
BUFFERED READS: 90.88 MB/sec
AVERAGE SEEK TIME: 8.20 ms
FSYNCS/SECOND: 1144.69
DNS EXT: 36.46 ms
DNS INT: 10.87 ms (ovh.net)
root@fr:~#


i7-2T kimsufi

root@fr :~# pveperf
CPU BOGOMIPS: 42562.16
REGEX/SECOND: 408928
HD SIZE: 19.38 GB (/dev/md1)
BUFFERED READS: 86.24 MB/sec
AVERAGE SEEK TIME: 9.92 ms
FSYNCS/SECOND: 669.22
DNS EXT: 32.12 ms
DNS INT: 1.67 ms (ovh.net)
root@fr:~#

When i try to see what hardware is on the both machines (lshw |more) The i7-2T kimsufi takes really long to
print info. The Q-1T kimsufi is really fast ? A simple software install takes real long compare to the Q-1T kimsufi.
Looks like performance with harddrive ?