[pve-devel] adding libiscsi to qemu-kvm ?

Alexandre DERUMIER aderumier at odiso.com
Tue Jul 10 08:51:36 CEST 2012


>>Yes, that is why most people use LVM on top of iscsi. 
>>
>>Note: Simple ignore my ignorant comments ;-) I know that this does not help you. 

Yes, I would like to use lvm, but no thin provisionning, too much snapshot overhead,... :(
Maybe in the future redhat will put dm-thinp cluster aware ? 

>>Ok, please send the patch. Adding another storage plugin does not harm - 
>>I guess it is just a few lines of code? 
Yes, sure, very small plugin :)   (Network plugin architecture is very good :)

I'll polish the plugin this week and send it after :)


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

De: "Dietmar Maurer" <dietmar at proxmox.com> 
À: "Alexandre DERUMIER" <aderumier at odiso.com> 
Cc: pve-devel at pve.proxmox.com 
Envoyé: Mardi 10 Juillet 2012 08:39:05 
Objet: RE: [pve-devel] adding libiscsi to qemu-kvm ? 

> >>For me this is a step back. 
> I disagree.This is a huge step for direct iscsi luns. 
> I can tell you that it's currently a pain to manage with a lot of luns. 

Yes, that is why most people use LVM on top of iscsi. 

Note: Simple ignore my ignorant comments ;-) I know that this does not help you. 

> >>We lose snapshot capabilities and multipath support? 
> 
> which snapshot capabilities ? I'm talking about iscsi devices, not lvm on top 
> of iscsi. 

Ah, OK. 

> Snapshots/cloning must be done on the san side. (nexenta,netapp,.. have api 
> to do this). 
> About multipath, I can tell you that with 500luns, it's pretty impossible to 
> manage. (it's using a lot of cpu and hanging sometimes) In fact,i'm going to 
> implemenent lacp soon, I have add too much problem with multipath 
> daemon. 
> 
> 
> Currently if something goes wrong (failover of san controller or network 
> saturation), with a lot of luns, kernel take a lot of time to restore device 
> access. 
> last time, I have a 30sec failover, kernel+multipath have take 5min to restore 
> device access... 
> With libscsi, it take around 5sec to restore access. 
> >> And as soon an ceph/sheepdog is stable you will dump iscsi anyways? 
> 
> Sure, but I don't want to put to trash a 100.000$ san, right ? ;) 

Ok, please send the patch. Adding another storage plugin does not harm - 
I guess it is just a few lines of code? 

- Dietmar 




-- 

-- 



	

Alexandre D e rumier 

Ingénieur Systèmes et Réseaux 


Fixe : 03 20 68 88 85 

Fax : 03 20 68 90 88 


45 Bvd du Général Leclerc 59100 Roubaix 
12 rue Marivaux 75002 Paris 



More information about the pve-devel mailing list