>From my (production) experience from few years ago, even GFS2  as a
clustered file system was SOOOOOOOO unstable, and lead to locks, causing
the share to become unresponsive 100%, and then you go and fix the things
any way you can (and this was with only 3 nodes !!! accessing the share
with LIGHT write IO)...all setup by RH themselves back in the days (not
cloud, some web project)

Avoid at all cost.

Again, I need to refer to the work of Mike.Tutkowski from Solidfire, where
he implemented "online storage migration" from NFS/CEPH to SOLIDFIRE - I
hope that someone (DEVs) could implement this globally between NFS storages
for begining, if there is any interest, since you can very easily do this
migration manually with virsh and editing XML to reference new volumes on
new storage, then you can live migrate VM while it's running with zero
downtime...(so zero shared storage here, but again live migration works
(with storage))

Cheers

On 19 February 2018 at 15:21, Dag Sonstebo <dag.sonst...@shapeblue.com>
wrote:

> Hi Swen,
>
> +1 for Simon’s comments. I did a fair bit of CLVM POC work a year ago and
> ended up concluding it’s just not fit for purpose, it’s too unstable and
> STONITH was a challenge to say the least. As you’ve seen from our blog
> article we have had a project in the past using OCFS2 – but it is again a
> challenge to set up and to get running smoothly + you need to go cherry
> picking modules which can be tricky since you are into Oracle territory.
>
> For smooth running I recommend sticking to NFS if you can – if not take a
> look at CEPH or gluster, or ScaleIO as Simon suggested.
>
> Regards,
> Dag Sonstebo
> Cloud Architect
> ShapeBlue
>
> On 19/02/2018, 13:45, "Simon Weller" <swel...@ena.com.INVALID> wrote:
>
>     So CLVM used to be supported (and probably still works). I'd highly
> recommend you avoid using a clustered file system if you can avoid it. See
> if your SAN supports the ability to do exclusive locking.
>
>     There was chatter on the list a couple of years ago about a scaleio
> storage driver, but I'm not sure whether there was any movement on that or
> not.
>
>
>     - Si
>
>
>     ________________________________
>     From: S. Brüseke - proIO GmbH <s.brues...@proio.com>
>     Sent: Monday, February 19, 2018 4:57 AM
>     To: users@cloudstack.apache.org
>     Subject: KVM with shared storage
>
>     Hi @all,
>
>     I am evaluating KVM for our Cloudstack installation. We are using
> XenServer at the moment. We want to use shared storage so we can do live
> migration of VMs.
>     For our KVM hosts I am using CentOS7 with standard kernel as OS and
> for shared storage I am evaluating ScaleIO. KVM and ScaleIO installation is
> working great and I can map a volume to all KVM hosts.
>     I end up with a device (dev/scinia) of block storage on all KVM hosts.
>
>     As far as I understand I need a cluster filesystem on this device so I
> can use it on all KVM hosts simultaneously. So my options are ocfs2, gfs2
> and clvm. As documented clvm is not supported by Cloudstack and therefore
> not really an option.
>     I found this shapeblue howto (http://www.shapeblue.com/
> installing-and-configuring-an-ocfs2-clustered-file-system/) for ocfs2,
> but this is for CentOS6 and I am unable to find a working ocfs2 module for
> CentOS7 standard kernel.
>     [http://www.shapeblue.com/wp-content/uploads/2017/03/
> Fotolia_51644947_XS-1.jpg]<http://www.shapeblue.com/
> installing-and-configuring-an-ocfs2-clustered-file-system/>
>
>     Installing and Configuring an OCFS2 Clustered File System ...<
> http://www.shapeblue.com/installing-and-configuring-an-
> ocfs2-clustered-file-system/>
>     www.shapeblue.com
>     Last year we had a project which required us to build out a KVM
> environment which used shared storage. Most often that would be NFS all the
> way and very occasionally ...
>
>
>
>
>     So my question is how are you implementing shared storage with KVM
> hosts in your Cloudstack installation if it is not NFS? Thx for your help!
>
>     Mit freundlichen Grüßen / With kind regards,
>
>     Swen
>
>
>
>     - proIO GmbH -
>     Geschäftsführer: Swen Brüseke
>     Sitz der Gesellschaft: Frankfurt am Main
>
>     USt-IdNr. DE 267 075 918
>     Registergericht: Frankfurt am Main - HRB 86239
>
>     Diese E-Mail enthält vertrauliche und/oder rechtlich geschützte
> Informationen.
>     Wenn Sie nicht der richtige Adressat sind oder diese E-Mail irrtümlich
> erhalten haben,
>     informieren Sie bitte sofort den Absender und vernichten Sie diese
> Mail.
>     Das unerlaubte Kopieren sowie die unbefugte Weitergabe dieser Mail
> sind nicht gestattet.
>
>     This e-mail may contain confidential and/or privileged information.
>     If you are not the intended recipient (or have received this e-mail in
> error) please notify
>     the sender immediately and destroy this e-mail.
>     Any unauthorized copying, disclosure or distribution of the material
> in this e-mail is strictly forbidden.
>
>
>
>
>
> dag.sonst...@shapeblue.com
> www.shapeblue.com
> 53 Chandos Place, Covent Garden, London  WC2N 4HSUK
> @shapeblue
>
>
>
>


-- 

Andrija Panić

Reply via email to