FreeNAS & TrueNAS Plans - 2020 and Beyond!

JoeAtWork

Contributor
Joined
Aug 20, 2018
Messages
165
This is critical. I'm not interested in running my servers on Linux. I choose FreeBSD for my servers for a reason (router, NAS, and other servers including internet-facing ones).

I hope that you never lose the FreeBSD version, and that these other efforts don't dilute the resources available to give due attention to the FreeBSD-specific builds. I can think of other projects that, after spreading themselves more thinly, basically gave the finger to their FreeBSD version. So my concerns are based on real life and not hyperbole.

I agree. FreeNAS based on FreeBSD we can run and not worry about all the issues we have with Linux and Windows. It is truly a breath of fresh air in the world of computing.
 
Last edited:

Lobout

Cadet
Joined
Jul 26, 2016
Messages
1
FC is an enterprise feature, we don't normally see it deployed outside of an enterprise environment.

if that is the case then you dont really know your user base very well. I have been running freenas with the FC hack since freenas 8. I would personally love to see FC features made available to us "Home" users.
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
PF is Packet Filter, the FreeBSD firewall that comes from OpenBSD folks.

Yeah, um, no. PF was sourced from OpenBSD a long time ago and has since drifted substantially.

FreeNAS already supports the more mainstream "ipfw" firewall system. I don't really suggest that you rely on an on-FreeNAS firewall, as FreeNAS defaults to allow, so a failure of your firewall can go unnoticed.

And fail2ban, while very popular with the Linux crowd, is dependent on Python and tailing logfiles. SSHGuard does all the heavy lifting in C, and can be fed from stdin via syslogd, which means that there isn't a possibility that a disk full condition could cause you to fail2ban. (See what I did there? Haha. Truth in naming!) On the other hand it does introduce a dependency on syslogd. No perfect solution.
 

Patrick M. Hausen

Hall of Famer
Joined
Nov 25, 2013
Messages
7,740
On the other hand it does introduce a dependency on syslogd.
I have yet to see a FreeBSD system without syslogd ...

Patrick
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
I have yet to see a FreeBSD system without syslogd ...

syslogd can be teased into crashing in a variety of ways. However, most data flows people actually use for fail2ban/sshguard pass through syslogd anyways, so this is really a question of two significant ways to muck up the works (fail2ban, syslogd dies or disk fills) or one significant way (sshguard, syslogd dies).

Also worth noting that the default FreeBSD install of sshguard tails the logfiles instead of taking a feed directly from syslogd. (rolls eyes).
 

seanm

Guru
Joined
Jun 11, 2018
Messages
570
As soon as ZFS gets the all clear I can see where all UNIX type OSes should default to use it, I can also see Uncle Larry wanting to license it to Microsoft as they have rubbish file systems and Apple as well.

Apple almost switched to ZFS year ago, but for various reasons didn't. You can argue HFS is rubbish, but APFS is pretty nice. ZFS is great and all, but would it really be the best choice for a constrained device like the Apple Watch? Apple didn't think so apparently.
 

brando56894

Wizard
Joined
Feb 15, 2014
Messages
1,537
The FreeNAS UI on Linux would be pretty damn cool. I found FreeNAS back in the days of 9.3 and loved the UI and ZFS, but wasn't really a fan of FreeBSD since I had been using Linux for almost a decade. I kept switching back and forth between FreeNAS 9.3, 10, unRAID, OpenMediaVault, Proxmox and CentOS w/Cockpit as my base OSes. Then I switched to ESXi and had FreeNAS handling my storage, and Linux handling everything else, but in the end I always went back to just Linux. I'm currently using Arch Linux w/ZoL to do everything, via the CLI mostly.

I created my zpools in Linux about a year or so ago so I've been stuck with Linux since then, and I'm anxiously awaiting v12 and the OpenZFS merger so I could split everything again with a type 1 hypervisior, but if I can run the UI on Linux I wouldn't have to split it!
 

RegularJoe

Patron
Joined
Aug 19, 2013
Messages
330
Two ideas:
FreeNAS as iocage jail
FreeNAS compartmentalized to be like a hyper-visor with SDN and data/control/service/management planes

I know this sounds crazy but it would allow us to reboot FreeNAS 1 second or a few seconds.... think iSCSI or NFS server for VMware having a non-disruptive boot.
 

nikalai2

Dabbler
Joined
Jan 6, 2016
Messages
40
Very good ideea and from what i know, there were rumors regarding HW transcoding. :)
 

RegularJoe

Patron
Joined
Aug 19, 2013
Messages
330
GPU pass-through?

That would be a bhyve limitation and when bhyve supports it and we get that version in FreeNAS we can do it.
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
Very good ideea and from what i know, there were rumors regarding HW transcoding. :)

Thats not rumors, thats "A lot of people not knowing how to do it and what the current limitations are" ;)
I can verify this guide to do it works on 11.3 already for a limited subset of intel cpu's:
 

danb35

Hall of Famer
Joined
Aug 16, 2011
Messages
15,456
I still just want to be able to expand a vdev by 1 disk
That's a ZFS limitation. There's been work in that direction, but it appears (based on the fact that it was expected some time ago, but has yet to be released) that the project was trickier than had been expected. I don't know the current status other than that it isn't released yet.
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
That's a ZFS limitation. There's been work in that direction, but it appears (based on the fact that it was expected some time ago, but has yet to be released) that the project was trickier than had been expected. I don't know the current status other than that it isn't released yet.

It also has to do with resources...
The maintainers of ZFSonLinux (which are also the most active developers) barely have the time to properly review new features already, which tells us something about how much time they have to write low-priority but high-complexity features themselves.

It's a bit of triage.
If they work on complex features, they also need to take into account to userbase and, sadly enough, homelabbers who want to add a single disk, is not the highest priority. For example draid is also a feature thats slowly crawling towards completion, but als has a HUGE usecase for corporate users.

It might also be some devs holding off on it, because DRAID and the ZoL-ZoLoF merger is going to shake up the market/codebase pretty badly...
 

danb35

Hall of Famer
Joined
Aug 16, 2011
Messages
15,456

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
This feature was presented by Matt Ahrens (another discussion thread here: https://www.ixsystems.com/community/threads/raidz-expansion-its-happening.58575/), and was said to be moving along nicely code-wise, so I don't think it can be said to be a lack of interest among the devs. But no doubt the rebasing of OpenZFS to use Linux as the base implementation is adding complications.

Hold on:
I never said it was a lack of interest. I said it was a lack of TIME and developers.

I think @Arwen explained it quite nicely:
"So it is not dead, just taking time due to other things of higher priority. (Like the FP/SIMD save/restore for RAID-Zx calculations.) "

Thats not a "lack of interest" thats a lack of time and developers.
And yes, priorities DO mater, but are not the primary reason things get on a backburner. Just the fact one doesn't have the time, doesn't make him more or less interested in a project ;)


Please just forward to the actual github issues next time, instead of this forum. Not offence, but this forum is not in any way, shape or form a relevant discussion platform for the actual state of current dev on ZoL. https://github.com/openzfs/zfs/pull/8853

The TL:DR of that thread is: "there is discussion about it, but no one is currently actively writing the code"


You seem to not-know the fact that most ZoL big features have a development timeframe of about 2-4 YEARS after the first time they are demonstrated and stated to be "moving along nicely".

Look at Draid, vdev expansion, vdev removal, zstd etcetcetc.
 

Patrick M. Hausen

Hall of Famer
Joined
Nov 25, 2013
Messages
7,740
It's OpenZFS, not ZoL and the FreeBSD and Linux developers are working together on finishing the unification and stabilization of the code base. Allan Jude made a few valid remarks about vdev topology changes and why this feature although in high demand is probably last on the list. Listen to the latest episode of BSD now for that: https://www.bsdnow.tv/340

Second, the monthly OpenZFS leadership calls are public. Not in terms of participation, but everyone is invited to listen in or watch later, e.g. https://m.youtube.com/watch?feature=youtu.be&v=FVwYAwrKZCU

I may be a bit overly critical, but some of the points in these threads that come up time and time again do look like FUD to me. So please go to the sources if you have doubts about the future development of ZFS. As far as I am concerned I trust the people some of whom I know personally for quite some time now that the future of ZFS on FreeBSD is bright and probably better than ever.

Pool checkpoints and special allocation class vdevs are way more important to me, for example, than adding and removing single drives. Allocation classes are huge, folks! Next up: persistent L2ARC!

Patrick
 

jgreco

Resident Grinch
Joined
May 29, 2011
Messages
18,681
Please just forward to the actual github issues next time, instead of this forum. Not offence, but this forum is not in any way, shape or form a relevant discussion platform for the actual state of current dev on ZoL.

I'd suggest not being like that about it. iXsystems has been a primary driving force behind ZFS on FreeBSD for many years, and just because the underlying codebase for OpenZFS is changing to ZoL does not invalidate the contributions that have been made by FreeBSD over the years. Without FreeBSD having taken the lead in integrating and developing ZFS for many years, it is likely that the community wouldn't have had such exposure to ZFS, and it's quite likely that ZoL would not have come to pass.

There is no need to be disrespectful in this manner towards the forum. If you don't think the forum is relevant, please feel free to troll elsewhere.
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
It's OpenZFS, not ZoL and the FreeBSD and Linux developers are working together on finishing the unification and stabilization of the code base.

I don't think I need educating on a project I also work on myself.
During the time Matt made the remarks here where danb35 refered to it was still ZoL and most of the features talked about here started dev during the ZoL time period. I made the choice to use ZoL for a reason.

Allan Jude made a few valid remarks about vdev topology changes and why this feature although in high demand is probably last on the list. Listen to the latest episode of BSD now for that: https://www.bsdnow.tv/340

Cool, I never have time for podcasts... thanks for the reference!

Second, the monthly OpenZFS leadership calls are public. Not in terms of participation, but everyone is invited to listen in or watch later, e.g. https://m.youtube.com/watch?feature=youtu.be&v=FVwYAwrKZCU

They are open both in participation AND watching...
Click: https://docs.google.com/document/d/1w2jv2XVYFmBVvG1EGf-9A5HBVsjAYoLIFZAnWHhV-BM/edit
They are also announced in the mailing list.

I may be a bit overly critical, but some of the points in these threads that come up time and time again do look like FUD to me. So please go to the sources if you have doubts about the future development of ZFS.
I don't think you are overly critical, I scolded quite a few people myself for things like using ZoL to describe OpenZFS2.0
And yes, more people should just go to the source (github, meeting reports, mailinglist etc.) instead of refering to a post from 2017. Or Allan on BSDNOW is also a good source, mostly first hand.


As far as I am concerned I trust the people some of whom I know personally for quite some time now that the future of ZFS on FreeBSD is bright and probably better than ever.

Yes, this is very true... There are some awesome features planned to be added to OpenZFS (including bsd ;) ) this year... Think: Draid, ZSTD, Persistent L2ARC etcetc... This would give a HUGE boost to TrueNAS core and ZFS in general :)

Pool checkpoints and special allocation class vdevs are way more important to me, for example, than adding and removing single drives. Allocation classes are huge, folks! Next up: persistent L2ARC!

With a bit of luck, Behlendorf is considering it, persistent L2ARC should get into BSD together with Allocation Classes when OpenZFS2.0 drops... Don't underestimate ZSTD btw, if it makes it in OpenZFS2.0 that would totally change the compression landscape.

Also: I'm super hyped for the combination of DRAID and Allocation classes, the synergy between those is amazing

TL:DR:
Super awesome features comming, even without vdev expansion.
 
Top