TrueNAS SCALE Announcement and Nightly Image Downloads

usergiven

Dabbler
Joined
Jul 15, 2015
Messages
47
Data can be striped across nodes into a larger pool... (no HA in this case)
Thank you! What if a pool fails on one of the two striped hosts. Do you lose the data on both hosts or only the host with the failed pool? I'm thinking if the the grouping technology is Gluster then maybe it doesn't retain the ZFS issue there and you would only lose part of your data.
 

morganL

Captain Morgan
Administrator
Moderator
iXsystems
Joined
Mar 10, 2018
Messages
2,691
If you lose a pool within a cluster stripe you will effectively lose all the data in the stripe..... don't assume its usable for most applications.

However, most of the time, you won't lose a pool, just a node will fail. If the node can be later restored/replaced with its pool, then no data is lost in that pool or the stripe.
 

morganL

Captain Morgan
Administrator
Moderator
iXsystems
Joined
Mar 10, 2018
Messages
2,691
Any considerations for a witness-only role just to avoid split-brain?
Yes, the 3rd node can be the "arbiter" node without a full copy of the data. It does keep track of data that has changed and needs to be restored.
 

PackElend

Explorer
Joined
Sep 23, 2020
Messages
60
May I share my successful install of TrueNAS SCALE UDOO BOLT GEAR but I wouldn't call it stable.
  1. It recognizes the installed hardware (could create a zfs pool) but can't get detailed readings like temperature. I booted Ubuntu live image what was able to get all the readings
  2. Nextcloud app was not able to initialize due to account creation error with SQLite
  3. Launched portainer's Profile - Docker Hub but it was not able to get access to the Docker daemon, so I tried to repeat it adding the command
    Code:
    -v /var/run/docker.sock:/var/run/docker.sock
    in container arguments (args), what was properly the wrong place but it messed up anything. I was not able to get in running again, even with delete/reinstall.
    Deploying the docker container via shell work like as usual but the container is not listed in "Applications".
  4. When I go in "Application" I get numerous pop-ups that the pool cannot be read (what might be related to #1)
Doing all these checks etc. raised the question, how to reports this thing properly: I cannot find a clear description of how to this, forum, but where for SCALE) or https://jira.ixsystems.com/.
Is there a standard template, console output or log for troubleshooting?

I noticed that deploying the nextcloud application includes rancher-os and Kubernetes container. Is there any intention to use Kubernetes as a standard container deployment engine?

As docker is basically working but SCALE is UI in regard of docker ist not that useable I would try to have all containers' images, volumes etc located on zfs pool, what can be mounted into Ubuntu. That would allow me to do the same on Ubuntu to enable to run the same container on both OSs, switching easily between testing and production, as all my apps are containerized.
Any doubts/thoughts that theory could fail in practice?

Last but not least, is there any rough estimation of Alpha 21.02, there are not that many open issues open anymore.
 

morganL

Captain Morgan
Administrator
Moderator
iXsystems
Joined
Mar 10, 2018
Messages
2,691
I noticed that deploying the nextcloud application includes rancher-os and Kubernetes container. Is there any intention to use Kubernetes as a standard container deployment engine?

Last but not least, is there any rough estimation of Alpha 21.02, there are not that many open issues open anymore.

Yes Kubernetes is the standard container deployment engine. The developer notes provide some basic info for this.
https://www.truenas.com/docs/hub/scale/dev-notes/

The 21.02 version is getting close to feature freeze... it can be accessed as a NIGHTLY (see developer notes) and should be available mid February.... quality permitting..
 

anodos

Sambassador
iXsystems
Joined
Mar 6, 2014
Messages
9,545
I just merged Samba 4.14.0rc1 into SCALE. Release notes are here: https://download.samba.org/pub/samba/rc/samba-4.14.0rc1.WHATSNEW.txt

Please feel free to ping me if you encounter a bug. Those of you who heavily customize via auxiliary parameters should take note of the following:
Code:
smb.conf changes
================

  Parameter Name                     Description                Default
  --------------                     -----------                -------
  smb encrypt                        Removed
  ldap ssl ads                       Removed
  client plaintext auth              Deprecated                 no
  client NTLMv2 auth                 Deprecated                 yes
  client lanman auth                 Deprecated                 no
  client use spnego                  Deprecated                 yes
  domain logons                      Deprecated                 no
  raw NTLMv2 auth                    Deprecated                 no
  async dns timeout                  New                        10
  client smb encrypt                 New                        default
  honor change notify privilege      New                        No
  smbd force process locks           New                        No
  server smb encrypt                 New                        default
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
May I share my successful install of TrueNAS SCALE UDOO BOLT GEAR but I wouldn't call it stable.
  1. It recognizes the installed hardware (could create a zfs pool) but can't get detailed readings like temperature. I booted Ubuntu live image what was able to get all the readings

  1. Thats unfortunate, but expected during alpha.

    [*]Nextcloud app was not able to initialize due to account creation error with SQLite
    To be frank, the Nextcloud app is awefull.
    And Nextcloud docker containers are my main example on how not to create a docker container.
    That being said: the Apps interface (and the apps) is mostly a Proof-of-Concept

    [*]Launched portainer's Profile - Docker Hub but it was not able to get access to the Docker daemon, so I tried to repeat it adding the command
    Code:
    -v /var/run/docker.sock:/var/run/docker.sock
    in container arguments (args), what was properly the wrong place but it messed up anything. I was not able to get in running again, even with delete/reinstall.
    Deploying the docker container via shell work like as usual but the container is not listed in "Applications".
    Non-K8S deplotments are not supported. They are possible but not actively supported. As are adding multiple k8s management utilities. You complain about stability, but try to costumise SCALE a lot (and hence go outside of the current development scope), which i think is not fair.

    That being said: Apps list SCALE apps, not all random docker containers. If you run your own docker container using docker-compose or plain docker, this is not a SCALE APP, hence is not listed as such

    [*]When I go in "Application" I get numerous pop-ups that the pool cannot be read (what might be related to #1)
    Might be, try rerunning the applications setup
Doing all these checks etc. raised the question, how to reports this thing properly: I cannot find a clear description of how to this, forum, but where for SCALE) or https://jira.ixsystems.com/.
Is there a standard template, console output or log for troubleshooting?
First off: Bugs can be reported on JIRA and under config you can output a debug dump.
However: Non-supported things not-working, are not bugs

I noticed that deploying the nextcloud application includes rancher-os and Kubernetes container. Is there any intention to use Kubernetes as a standard container deployment engine?
K8S IS the standard container engine. SCALE doesn't actually officially support non-k8s docker.
(As in: official support is not provided but it "should-work(tm)" )

As docker is basically working but SCALE is UI in regard of docker ist not that useable I would try to have all containers' images, volumes etc located on zfs pool, what can be mounted into Ubuntu. That would allow me to do the same on Ubuntu to enable to run the same container on both OSs, switching easily between testing and production, as all my apps are containerized.
Any doubts/thoughts that theory could fail in practice?
The SCALE UI only lists SCALE APPS, not containers or helm charts.
SCALE APPS are specially re-crafted helm-charts, that are managed by SCALE.

If you run the setup utility from the APPS UI, it makes sure all docker images are by-default put on the ZFS pool you select.
Going back-and-forward between different OS'es however, is not actively or offically supported

Last but not least, is there any rough estimation of Alpha 21.02, there are not that many open issues open anymore.
Actually there are quite a number of things that need working, some major design tasks are not merged yet. But indeed 21.02 should launch before the end of februari. The number of issues is often less important than the scope of said issues ;-)


Anyway:
In general I find it unfair you define the product as unstable, because it doesn't want what you want it to do.
You are doing things that aren't supported and complain things don't react like you expected. This is a prime example of layer 8 issues. Not the product.
 

PackElend

Explorer
Joined
Sep 23, 2020
Messages
60
In general I find it unfair you define the product as unstable, because it doesn't want what you want it to do.
was not my intention, it mainly referred to the fact that it all values are 0 in the hardware monitor and that I got that pop-up "cannot read pool" (or similar, cannot reproduce at the moment, as KS service is down). So I was afraid that there is a problem, using my hardware, so it could be that it runs unstable on my device.
It got quite late last night, as I posted it, will try to be more specific next time.

thanks for the comprehensive feedback, will try to revert back later or tomorrow.

Greate work so far, I'm exited of the first beta
 

horizonbrave

Explorer
Joined
Nov 15, 2016
Messages
56
May I share my successful install of TrueNAS SCALE UDOO BOLT GEAR but I wouldn't call it stable.

Please keep us updated on your progress, I might be interested in buying few boards for TrueNAS Scale use.
Your English looks a bit funny I presume you must be one the Italians involved with the project ;)
Please make sure to find a way to report the issues on Jira so that they can tackle it and hopefully find a solution which will benefit both projects A LOT (supported ECC ram in such a x86 small board is a game changer).
Thank you all!
 

horizonbrave

Explorer
Joined
Nov 15, 2016
Messages
56
However: Non-supported things not-working, are not bugs

What is this supposed to mean?
Why is that board not supported??
You guys should partnered up, it's such a SHAME that iXsystems machines are not available overseas (maybe they are, but current shipping costs make it unworth it). And this things are small and easy to ship?
I hope someone at iXsystems will take note of the suggestion ;)
 

horizonbrave

Explorer
Joined
Nov 15, 2016
Messages
56
Planning to make this functionality easy to use on Feb release. 21.02

I predict you'll have LOTS of people (ProxMox users) giving it a try / test (which is good for feedback).. please make sure to document everything crystal clear so to avoid lengthy and repetitive posts on the forums :)
 

Patrick M. Hausen

Hall of Famer
Joined
Nov 25, 2013
Messages
7,740

PackElend

Explorer
Joined
Sep 23, 2020
Messages
60
I did a fresh install of TrueNAS-SCALE-20.12-ALPHA:
  1. no error message when Kubernetes was initiated but no application running to confirm 100%
  2. hardware monitoring is working
  3. change of time zone was immediate
There is only:
ghlight_offgluster.peer.status 0.00%
Status: FAILED
Start Time: 25 Jan 2021 14:46:56
Finished Time: 25 Jan 2021 14:46:57
Error: [EFAULT] Connection failed. Please check if gluster daemon is operational.
---> does that need to be reported?

It was not just a day to read up certain things I may still need to continue. I would like to switch from Ubuntu to TrueNAS, so I asked if container could be used by both OSs.
Detailed questions in this regard in Trying to understand Kubernetes implementation and how to use in multiple OSs | TrueNAS Community although it looks like using the same What Is Etcd and How Do You Set Up an Etcd Kubernetes Cluster? (rancher.com) on all OSs is not trivial.

Moreover, I'm wondering, if there will be the support of virtual disk without then need to convert them in raw, at least truenas/pkg-qemu: https://salsa.debian.org/qemu-team/qemu (github.com) raised some hope.
More details in my answer on TrueNAS SCALE - Virtualization Plugin - File/qcow2 support for QEMU/KVM instead of using zvol | TrueNAS Community.

I would love to post frequently how it works with the BOLT, I could even do certain tests if there is a demand for it. I could imaging to run the latest release in parallel to the latest nightly if things can be shared between both easily.

(p. s. my very first post was done very late, but it looks like it still understood :smile:)

---- update

can confirm that the Nextcloud apps works flawless, cannot tell what made it fail during first install.
 
Last edited:
Joined
Feb 6, 2021
Messages
18
Greeting, everyone.
I played with TrueNAS-SCALE-20.12-ALPHA and updated it to the latest nightly version of TrueNAS-SCALE-21.02-MASTER-20210205-232916. I want to run some tests using iperf & fio, but when I type these commands in shell, I get a "command not found" message. Am I doing something wrong?
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
iperf and fio are not packages that have to be included in the OS, as they are not part of the kernel, plain debian nor part of the IX installer.
Manual install of those using apt-get update && apt-get install should work fine in this case.
 
Top