Hold up?!
Incus is getting removed?
2 Likes
Yes
I’m only reading that’s being “evaluated”:
Can TrueNAS make a competitive, well-featured virtualization function without harnessing/leveraging the work already in Incus? Wouldn’t that be the most efficient path?
I certainly don’t agree that Incus “failed” just because a few people couldn’t migrate old Bhyve VMs to a new platform.
I likely don’t agree that choosing different strategies over time is some big problem either.
“Release Early, Release Often”
“Many Eyeballs Tame Complexity”
It seems like many people here are NOT well adapted to open source methodology and think something can be better managed through hierarchical, authoritarian planning (with, of course, them calling the shots.) However, central planning doesn’t necessarily deliver the order it promises:
"[Conventionally-managed development doesn’t necessarily deliver] …reliable execution by deadline, or on budget, or to all features of the specification; it’s a rare `managed’ project that meets even one of these goals, let alone all three.
– http://www.catb.org/~esr/writings/cathedral-bazaar/cathedral-bazaar/ar01s12.html
Much of the “you should have done in this way…”/“I would have done it like this…” gripes here sound more like backseat driving tier bossiness.
Solutions evolve over time. Embrace each new possibility, figure out the puzzle before you, and build something cool with it.
Corral was neat – it had a fresh new UI and probably taught some lessons that went into the apps today.
K3s was cool – I learned about operators and Helm charts, and had AWX running on it for a while, as well as other stuff.
Incus was great – the basics of VM and containers covered in the UI – and tons of API-accessible extra power under the hood was amazing really.
The planned Libvirt LXC will probably have some great potential too.
Focus on Learning, Not Punishment
Celebrate Effort and Progress
Embrace Failure as a Prerequisite for Innovation
Encourage Risk-Taking (Calculated Risks)
Celebrate Experiments (Regardless of Outcome)
I just spent the better part of 4 months getting a PoC going and entire thread about this project and now I have to back everything back out? I was told this was going stable and moving forward…
This is me by the way.
2 Likes
Experimental
They haven’t in 15 years, what makes you think they care about doing so now?
What’s really frustrating is that these aren’t dumb people; they’re very smart. But they’re making lots of really dumb decisions.
1 Like
It’s bad enough my bug reports getting closed out. I’m out. Moving on to my own self build with Proxmox or Incus on Debian.
Apps suck on TNCE, jailmaker will likely not work anymore in the future. So there isn’t many places to go with the feature set of TNCE.
Ok, so just everyone knows my source of this info about end of Incus in Truenas.
- My question answered in T3 podcast: https://youtu.be/IV14vybHVWc?t=339
- And because it wasnt clear to me if they mean temporary removal or permanent, I also got clarification from kris in forum chat that its really total permanent removal. Message #6961 by kris – #General
4 Likes
It’s hard to make a business case to focus more developer attention on Apps and VMs when we’re told no Enterprise customers take an interest in same. As @Protopia pointed out, Enterprise customers with heavy-duty needs will dedicate machines to that task, just as they dedicate TrueNAS file servers to file serving.
I’d argue that more Enterprise customers could become interested in VMs and perhaps even some Apps IF the feature-set was better supported, easier to set up, and easier to maintain; especially if the Enterprise customer has an interest in staying somewhat current re: SCALE. I suppose this is where the support contract monetization cames back into play…
Anyhow, I’ll be really bummed when and if Jailmkr stops working as it’s been so helpful with @dan’s SSL deployment scripts.
1 Like
I get all that, I think the bigger issue is that IX doesn’t know what it wants to be. SCALE was meant to be hyperconverged with virtualization compute and storage support all in one. Now it’s back to just a storage box. That’s why I jumped over to SCALE a long time ago because I liked the ability to scale out my lab over time. There are other storage options that can be had to accomplish similar results to what Gluster did.
I think that’s why it’s more frustrating than anything. Where are we going here? What’s the plan?
I’m not minimizing the improvements they’ve brought to ZFS and the work they put in there.
2 Likes
That final comment, tbh, didn’t help iX’s position one bit. Dripping with sarcasm and deliberately misunderstanding and misrepresenting the points being made, then closing the thread. Shows that this is a contentious issue, if anything.
Because it makes sense. Because they said so.
Kris Moore: …let me just get on my soap box for two seconds about that so folks who’ve been paying attention when we did Scale the VM section has been the last section I’ve wanted to go and do kind of a massive overhaul to for a long time. It was essentially a fork of what we had with Core on using bhyve as the virtualization. It hadn’t changed much so this has been a long time coming and being able to move it to Incas I think is really going to set the lay the foundation for a lot of cool stuff that’s going to happen in TrueNAS around virtualization, containerization, LXC stuff. Again I’ve wanted to do for years. We just haven’t been able to get to it because we had other more pressing fires, but again it’s a good thing everyone should be excited. I think it’s going to change a lot over the next couple years.
Stéphane Graber: yeah and the and the user experience is is pretty good in that like you get to access pre-made images – for just about any Linux workload, you can basically just pick up pre-built thing and you’re just done. Like, you don’t need to build an ISO, attach it, install it all that kind of stuff. You still need to do that for things like Windows but for anything else it’s it’s a lot easier.
Kris Moore: …just plug and play just click and go it’s very nice so again I’m looking forward to getting the release out here we’re just a few days away again try it out we’d love feedback we are still calling it experimental on this release because we do have a lot more we want to do to it to kind of harden it and make it more enterprisey for our release in the fall so uh stay tuned i’m sure we’ll be talking about that in the future.
– https://www.youtube.com/watch?v=_wucXs0zYMM&t=2773s
But, more because it makes sense, from what I know (which is limited). However, if they can approach the feature set I listed above for VM/LXC via LibVirt controls, that’d be awesome too.
PrivatePuffin mentioned that Incus is coded with “strong opinions” as to handling storage, which opinions do not match those expressed in TrueNAS.
Irrespective of any philosophical issues, integrating two complex pieces of software was likely to be complex. The problem here was probably less about Incus than about TrueNAS (the company) failing to properly prototype the planned change.
And, in any case, plainly removing an existing feature without a fully featured replacement was a mistake of Corral proportion. Full stop.
Libvirt-managed VMs should have been retained alongside Incus. 25.04.2 will show that it was indeed possible. (If this was only found out after the backlash, see “failure to properly prototype” above.)
This looked at least like the most efficient path, including to the-company-previously-known-as-iXsystems. So by throwing out the baby with the smelly bathwater, they have implicitly commited themselves to developing a fully featured VM manager. Which is likely to involve even more work than integrating an existing fully featured VM manager.
As far as I am concerned, the mission is completed with flying colours!
I agreed to your post until this part. Because if all of this is developped not only without any extra revuenue for TrueNAS (which I accept) but also without any expectation that it will ever generate extra revenue, then the only reasonable advice is: DROP THIS ALL!
No VMs, no LXC. Possibly not even docker. Focus on storage, and only on storage.
Of course, to develop a robust and trustworthy virtualisation platform will take time and efforts—and then to prove the “trustworthy” part to potential business users will take even more time and efforts.
This cannot be achived by setting aspirational goals on impossibly tight and arbitrary time schedules. (“I do not think you can succeed”… for those who know Cosmic Encounter.)
3 Likes
Maybe it’s my poor English skills, but IMO this article made some very debatable statements.
Completed? Nah, it can always get worse.
ESR is controversial, to say the least.
2 Likes
I’d focus on what is delivered, not that which is said, especially if there has been a long history of badly-supported Apps going back to FreeNAS 9. VMs worked well enough in CORE but have undergone multiple deep revisions in SCALE.
For those of us for whom rebuilding VMs takes a long time, such changes can be aggravating. Especially if it happens repeatedly.
1 Like
Finding a definition of this abbr. is giving me a really hard time. What is it?
iX are almost at the point where I’ll take anything they say as evidence of its opposite. Not quite, but they’re trying hard to get there.
And what you’ve quoted, in fact, illustrates this. That quote is all about how Incus is going to make for this great virtualization experience; now he’s saying Incus is gone, never to return.
2 Likes
The author of that page (and site), Eric S. Raymond.
1 Like
I had this suggestion but wasn’t able to find the author’s name.