Can't be everywhere And this seems less fancy that communicating with KVM directly. Sadly, last time I inspected the "state of KVM APIs", it was far from XAPI features (in an agentless point of view)

It's hard to understand because sometimes there is a difference between being a tech person mastering a technology and being able to understand users need.

@fateknollogee The need for UI in Open Source is ultra-important, and I don't get why people don't get it.

It's even possible to make some money with that So when I heard someone telling this, it's really funny and remind me some disdain I met in the past when I started Xen Orchestra.

@francesco-provino said in Manage KVM through Cockpit:

Why everybody seems to need a GUI for KVM? Virsh can do absolutely anything, in a much more concise, fast and elegant way. It's also very bandwidth friendly.

That made me laugh a lot

1. Doc is not completed because the product isn't. Consider it as a draft.
2. Blog post is still a draft

@dustinb3403 It won't be XOSAN to the community. XOSAN is like XOA, a turnkey packaged solution.

You'll be able to mount a gluster volume when we'll open the Gluster driver (which allows to create a SR of Gluster type in short). Then you can do whatever you want with that, but you'll need to monitor it yourself, and understand how Gluster works. Something you won't require by using XOSAN

Also a trial version of XOSAN will be available for people who want to test it and bench it quickly (literally in few clicks)

Not a published blog post (still in draft) but here is some bench done, please follow this link and keep it for yourself

We hope to make it for the end of the month (phase II) with a robust health status and flexibility (XOSAN full diagnostic, allow to replace in one click any faulty node, and everything auto heal itself).

Also, we'll try to release the Gluster driver in Open Source for those who want to connect XenServer to an existing Gluster or create their own XOSAN manually.

XOSAN will be an extra product of Xen Orchestra (but you can purchase even in XOA Free), including the turnkey side (and believe me it's not trivial) with full diag status and health + flexibility, including pro support on it obviously.

We'll try to make a Phase III before the official release, with tiering capabilities (HDD + SSD cache to get best of both world).

We started to create the official doc here: https://xen-orchestra.com/docs/xosan.html

Remember the goals of XOSAN:

• protect your data thanks to replication of data on multiple (from 2 to 16) hosts
• provide XenServer high availability without buying a NAS or a SAN
• give you flexibility to grow your storage by adding new nodes
• work on all kind of hardware, from HDDs to SSDs
• fully monitored solution

# python -V
Python 2.7.5

On XenServer 7.1

It's completely free and easy to use XOSAN deployement process is complex, rely on various bricks and linked to xen-orchesra.com, so that would have require a lot of effort to make it available directly from the sources in a usable way.

But if you want to do it yourself, it's also possible:

1. create a VM in each host of your pool you want to have in the cluster
2. install gluster server in each, create the cluster
3. install gluster client in XS
4. modify the file SM driver into XenServer to fit with Gluster
5. restart the toolstack
6. mount it

This way, you could also bench it

It's available in XOA Free, so you can test it without restrictions

Just made some benchmarks recently: https://xen-orchestra.com/blog/new-xosan-benchmarks/

7.2 is here guys

https://xen-orchestra.com/blog/xenserver-7-2/

Calm down, it's not a big step from 7.1. In fact, it's the start of a more "rolling releasing" thing for 7.2 to 7.x+1 (where x > 1 obviously )

Next time XS will be upgraded (every 6 months approx) it will be available via simple patches.

7.1 is "frozen" (more or less patches) because it's an LTS.

@scottalanmiller Especially the streaming thing

Believe me, it wasn't a piece of cake, due to the extreme old age tooling in dom0 (curl from 2006 in XS 6.5! Doesn't support SNI etc.)

edit: we even had to write our own HTTP server…

@scottalanmiller said in Xen Orchestra news :

Sweet. So it installs to the Dom0?

Nope, that's a bad idea It downloads the virtual appliance, set stuff and boot it.

For those who are using XenServer and want to take a quick look at XO, now you just had to SSH into your host and:

bash -c "$(curl -s http://xoa.io/deploy)"  

That's it

Nothing is stored on the host local disks, it's streamed directly to the default storage while downloading, and you can set the IP before. When it's done, you just have to click on the link that will redirect you to XO web UI.

https://xen-orchestra.com/blog/one-line-to-deploy-xoa/

Hi there,

XOSAN beta is meant to be done on XOA only because it uses mechanism related to xen-orchestra.com (you need a valid account there, with a valid token). So it's not possible to try XOSAN via XO on the sources.

It's still in a really early phase right now, so before going further we'll validate all the Phases explained in the blog post.

About the StarWind thing, I don't know the product at all, but on our side it's:

• XenServer only
• Thin provisioned
• Pool wide, ie working from 2 nodes up to 16 (well in theory more, but it's pool-wide not across pools)
• Uses Duplication/Triplication but also Disperse mode, depending of your need
• Auto heal (when a host/node is coming back online, it fetches all the missing data)
• Easy install/deploy (thanks to XOA)
• And more (basically, everything you can do with GlusterFS but "in a box", so think about HDD+SSD caching, complex disks usage to avoid any hardware RAID etc.)

@brandon

Pricing include:

• the virtual appliance with everything in it (plus the appliance is secured and optimized)
• the web update to just click on "update" when we release (often). So no CLI
• our pro support for any XOA issue you could encounter (eg a big complicated network)
• all updates

@brandon In theory, a home network is far more simple. So the script may work (not mine, so I can't tell)

It's 15 days, but I can extend it if necessary.