@Stell said:

Yep! Go nuts :)

:-)

@Stell said:

we’ll be supporting PVE 9.2 on this release

Ok, so I am safe to up to 9.2.x and 1.2.0 will handle it? :)

Thanks :)

I really wanted it to match our new site as well as WHMCS so clients get a fully “brand” experience…

@Stell said:

That covers disk cache mode, IO threads, AIO mode, discard, SSD emulation, and the recommended SCSI controller.

Nice! Been waiting for this for awhile ;)

@Stell said:

White-label branding

There is only one login page at least in 1.1.3? How does one brand that? And does branding support multiple FQDNs based on which reseller themes load?

@Stell said:

Service Provider login isolation. Service Provider administration can now be restricted to its own dedicated login URL.

YES!!! CloudFlare WAF here we come! ;)

Overall - this sounds great, but I must second @cmitchell - Is 9.2 supported, as otherwise a good number of us can’t / should not upgrade to 1.2.0 if not.

Thanks! Good job team!

BTW, if/when you guys get this Tenant API going, Cloud Propeller will MIT-style open source our in-house MultiPortal CLI tool for all to use … ;)

Howdy,

I do not believe this currently exists anywhere, but given how Proxmox updates work in typical Linux fashion — especially with the Enterprise Repo, which most of us are using in production environments — apt-get applies the latest available updates without an easy way to choose a specific version.

Because of that, we need a dedicated page on the wiki, as well as a clear note in every release note entry — both on the wiki page and in the .md file shown inside MP — stating which Proxmox version that MP release was tested with and approved for.

In addition, I believe it is important to establish a clear contract/promise between MP and its customers where:

• At the time of release, every version of MP supports the latest Proxmox Enterprise-repo version.
  For example, MP 1.1.3 should have been tested against the current Proxmox 9.2.x release available at that time.

To be clear, this does not mean all new Proxmox features must immediately be supported. However, it does mean that all existing functionality that MP uses or relies on must continue to work.

This should also be relatively straightforward to test and re-test. I would expect that by now there is, or should be, an automated test harness that validates MP functionality after every MP patch or codebase update against the supported Proxmox version.

It is critical for us to be able to keep Proxmox fully up to date at all times without risking broken MP functionality. Proxmox and its API are mature and generally follow good practices, so in theory most updates should not break existing MP integrations. The recent HA mechanism change is a good example of the kind of exception where MP needed to adapt — but that should be rare.

That is why the first request above is so important: we need a solid, authoritative page that clearly states which MP versions support which Proxmox versions, so there is no ambiguity for production customers.

Hello,

I guess this is the intro forum so might as seed it with first post.

I am CEO & Chief Architect at Cloud Propeller. Also, a huge fan of MultiPortal.

Also, a happy contributor of not just bug reports, but solutions on how to fix them. Matt can attest to that :)

At Cloud Propeller we specialize in providing Enterprise Infrastructure as a Service and were (for 10+ years) exclusively a VMware shop, however, with all the Broadcom snafu, after a lot of testing, we decided it was time to move to Proxmox.

Thanks to sheer luck, MP came onto the scene just in time as without it we’d have to write the whole thing (and were almost on the way to doing so).

Our old/current site still references VMware (it will, however, be replaced with the new site in about a week). If anyone here is interested in our Platform Architecture or Network Architecture, please be my guest and check it out.

We do have quite a setup if I may say so (4 x 100 Gb NICs, 2 x 32 core fastest single-core Xeons on the market, 2.3 TB of DDR5 + 12 x 15 TB Kioxia NVMe SSDs for CEPH, per node).

Also, we have an “open kimono” approach to our infrastructure so we show it in detail (with pictures, logical diagrams, etc.) on our new site. If there are any other nerds here like us - you’ll probably enjoy the two pages I mentioned.

On another interesting note, we may also be the only MultiPotal SP out there with a filly customized/themed instance…

Lots of work (~50 hours) went into making theming possible in a way that all of our modifications sit on top of MP’s own git repo (as an orphan branch), so when it’s time for updates, MP’s hard git reset simply wipes it all out after which our deploy shell scripts re-deploy it on top of the new version :)

We also built our own WHMCS Billing and Provisioning module which we may turn into an actual product other MP users can subscribe to (if there’s interest).

What it originally started (2 months ago) as a git clone of MP’s own WHMCS module, is now 99.9% custom / re-written MP Server Provisioning and Billing module. We started improving upon it, until we rewrote it completely. About 300+ hours went into this alone.

Keen eyed among you may have noticed two extra tiles in my MultiPortal screenshot above - showing Unbilled Usage + Average Daily Spend for a PayGo client, wight within the MP VDC page…

Here’s a few screenshots on how it looks like. Cool features include: Full support for reseller accounts (where on signup clients are made resellers, and their vdc is deployed into their own reseller account) - kind of how cPanel did with “client is a reseller, owns own account” checkbox, no more 1:1 limitation of MP Server per client, global shared UUIDs and such, All pricing is moved into Configurable Options (for both PayGo and Allocation type DCs). 5-minute PayGo billing (live, billed data available to users and admins every 5 minutes), full whmcs client-side integration allowing detailed info to clients, etc,.

I do not know if Forum Image resizer makes these too blurry but you can see all screenshots here: https://mirror.az31.cloudpropeller.com/tmp/mp/

Well, with that I leave you … For the very first introduction I hope you won’t mind the length… After all, who cares who I am or where we are from. I personally always found “what you do” or “how you do it” usually gives a bit context.

Cheers…

This would be great.

I was too busy integrating things into billing and such that I failed to notice Tenant API was missing. My assumption was that it is there until I had to actually use it ;)

@Stell said:

we’re performing a ground-up rewrite of our authentication and authorisation system which will deliver excellent granularity and control over per-tier API endpoints

Just please do not break existing one for admins (i.e. no API regression, rather /v2 introduction possibly). We have a lot of billing and outside-of-mp-lane automation depending on this.

Thanks.