[PVE-User] NTFS/Windows Server corruption aftersuccessful livestorage migration

Chris Murray chrismurray84 at gmail.com
Fri Dec 19 11:31:22 CET 2014 

Just a quick follow-up on this, we're on "qemu-server: 3.3-3" now, and although it was never 100% reproducible, we haven't observed the same symptom since the upgrade. I notice the new 'busy' indicator during the disk move. :-)

Thank you!

-----Original Message-----
From: pve-user [mailto:pve-user-bounces at pve.proxmox.com] On Behalf Of Chris Murray
Sent: 19 November 2014 17:41
To: Alexandre DERUMIER
Cc: pve-user at pve.proxmox.com
Subject: Re: [PVE-User] NTFS/Windows Server corruption aftersuccessful livestorage migration

Thank you for that Alexandre. I'm not a developer, so I don't fully understand the code, but I suppose I understand the principle of it :-)

Since it's not always reproducible on production hardware, I'll build a new server in the coming days and hook it into the pvetest repository.

-----Original Message-----
From: Alexandre DERUMIER [mailto:aderumier at odiso.com] 
Sent: 19 November 2014 13:58
To: Chris Murray
Cc: pve-user at pve.proxmox.com
Subject: Re: [PVE-User] NTFS/Windows Server corruption after successful livestorage migration

>>Out of interest, can I find information on the bug that has been fixed? 

Well, the main problem that we rely only on the progress of the copy, then at 100% we switch disk.
But this is wrong, because it's possible to have pending writes, so we need to check another qemu flag.

Here the commit:

https://git.proxmox.com/?p=qemu-server.git;a=blobdiff;f=PVE/QemuServer.pm;h=85ce28ad29e3c2d3746a80c50fd5b3ca0f92bbd2;hp=a79606cdd953aee789a968fd6de7c00feaa6f155;hb=ab6ecffe53a1d045dc03d2f9cf56037a8a2a9b2a;hpb=5fe03fd8eebb0fcc8a70d89f3810535fb2d42ec9


----- Mail original ----- 

De: "Chris Murray" <chrismurray84 at gmail.com> 
À: "Alexandre DERUMIER" <aderumier at odiso.com> 
Cc: pve-user at pve.proxmox.com 
Envoyé: Mardi 18 Novembre 2014 10:10:55 
Objet: RE: [PVE-User] NTFS/Windows Server corruption after successful livestorage migration 

Hi Alexandre, 

I do, apologies, I use the term 'storage migration' because of my past familiarity with another vendor :-) 

I'll work toward making the problem reproducible first if possible and will then try, thank you. Out of interest, can I find information on the bug that has been fixed? I notice a mention of the same file in the thread "Online storage migration not working (II)". As I understand Eneko's problem, migrations were failing? Rather than mine where they appear to succeed but (possibly) yield corruption. 

Thanks again for the assistance, and for a great product! 

-----Original Message----- 
From: Alexandre DERUMIER [mailto:aderumier at odiso.com] 
Sent: 14 November 2014 14:04 
To: Chris Murray 
Cc: pve-user at pve.proxmox.com 
Subject: Re: [PVE-User] NTFS/Windows Server corruption after successful livestorage migration 

Hi, 

do you talk about "move disk" feature of proxmox ? 

If yes, please test 

http://download.proxmox.com/debian/dists/wheezy/pvetest/binary-amd64/qemu-server_3.3-3_amd64.deb 

We have fixed a bug with storage migration 

----- Mail original ----- 

De: "Chris Murray" <chrismurray84 at gmail.com> 
À: pve-user at pve.proxmox.com 
Envoyé: Jeudi 13 Novembre 2014 18:13:40 
Objet: Re: [PVE-User] NTFS/Windows Server corruption after successful livestorage migration 



Is it possible that a developer could comment on the mechanics of the online migration algorithm so that I know where to troubleshoot next? 

I did create a rudimentary test in Debian which would write alternate patterns of data while I migrated VMs back and forth. Between each pattern it will test the file to verify that it contains everything it should. This all seems to work fine and the problem isn’t reproducible. Could the problem be OS-specific? I’m not a developer but I would guess that the migration algorithm would be lower-level than this? 

Is it possible that the corruption could affect Windows machines while they’re under load and while they’re being migrated? It seems like it’s a factor of those three things. I don’t think it’s purely load as they do receive much higher load when e.g. at 03:00 all Windows virtual machines are subject to new updates. This doesn’t cause a problem; it only seems to be if I move a disk while it’s busy. Everything will tick along fine and the moment I have another corrupt VM I ask myself “what happened last?” and the answer is “I moved it” or “I changed file type”. 

Any ideas? 

Thanks, 
Chris 



From: pve-user [mailto:pve-user-bounces at pve.proxmox.com] On Behalf Of Chris Murray 
Sent: 05 November 2014 22:34 
To: Yannis Milios; pve-user at pve.proxmox.com 
Subject: Re: [PVE-User] NTFS/Windows Server corruption after successful livestorage migration 

Hi Yannis, 

I can try that, but what is it expected that either result will prove? 

If the VM is corrupt, what will that prove? What is the likely cause of the problem? 
If the VM is not corrupt, what will that prove? 

Thanks, 
Chris 


From: Yannis Milios [ mailto:yannis.milios at gmail.com ] 
Sent: 05 November 2014 20:44 
To: pve-user at pve.proxmox.com ; Chris Murray 
Subject: Re: [PVE-User] NTFS/Windows Server corruption after successful live storage migration 

hello, 
Never tried online storage migration but what happens if you do the following: 
1. create a windows vm in raw format, on local disk storage(not nfs mount). 
2. Start installing updates on windows and initiate online storage migration to the nfs mount in qcow2 format. 
Do you experience the same issue? 
referance: https://pve.proxmox.com/wiki/Storage_Migration 


No virus found in this message. 
Checked by AVG - www.avg.com 
Version: 2015.0.5315 / Virus Database: 4189/8510 - Release Date: 11/04/14 


No virus found in this message. 
Checked by AVG - www.avg.com 
Version: 2015.0.5315 / Virus Database: 4189/8510 - Release Date: 11/04/14 
_______________________________________________ 
pve-user mailing list 
pve-user at pve.proxmox.com 
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user 

----- 
No virus found in this message. 
Checked by AVG - www.avg.com 
Version: 2015.0.5557 / Virus Database: 4213/8565 - Release Date: 11/13/14 

-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2015.0.5577 / Virus Database: 4213/8585 - Release Date: 11/17/14
_______________________________________________
pve-user mailing list
pve-user at pve.proxmox.com
http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-user

-----
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2015.0.5577 / Virus Database: 4213/8585 - Release Date: 11/17/14

More information about the pve-user mailing list