Applications UI Lands in SCALE Nightly Images

dalnew

Dabbler
Joined
Dec 9, 2020
Messages
26
I like the idea of the helm charts and configuration and the new UI but I couldn't quite figure out how they link together. I am able to use the UI to get some basic docker containers up and running, but then I couldn't find any corresponding config files generated that represent the data I entered in the UI. I actually had to do that for plex because the basic one that's installed doesn't have things like hostpath volumes and when I tried to add environment variables it crashes (or rather the window disappears and it never deploys). The default configuration bit seems to work for me though.

I expected to see a set of helm charts and yaml files with the corresponding config, but I don't see those anywhere. I do see a set of yaml files under for my "plex" docker under
ix-applications/releases/plex/charts/2012.0.0

It seems to be the one I generated based on the name and timestamps (and it goes away if I delete it), except none of the config stuff I entered is present in any of those files. Is that stored somewhere else? The files that are there just seem to be some generic configuration. I would love to be able to modify those files offline and then redeploy in the UI but maybe that's not possible yet? Or do I need to use some specific CLI commands for that? The commands in the documentation just show deployment using long command line definitions.

Also while I managed to get the plex docker up I still haven't figured out to add support for nvidia within the container. I know I'm supposed to add something like:
"gpuConfiguration": {"nvidia.com/gpu": 1}} that I see in the yaml files but since I used the UI and don't have any corresponding yaml files I'm not sure how to set this. Or even how that's represented in the old style docker-compose.yaml files... which is what I was using prior to switching over to kubernetes/helm in the UI.
 

morganL

Captain Morgan
Administrator
Moderator
iXsystems
Joined
Mar 10, 2018
Messages
2,691
I expected to see a set of helm charts and yaml files with the corresponding config, but I don't see those anywhere. I do see a set of yaml files under for my "plex" docker under
ix-applications/releases/plex/charts/2012.0.0

Agree documentation is not (nowhere near) complete, but did you find this doc from the developers notes?

 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
Agree documentation is not (nowhere near) complete, but did you find this doc from the developers notes?

Not really helping, because the information most important to @dalnew is totally missing from there.

@dalnew
I agree the documentation is limited, so lets go over it:
- IX Apps are made from helm-chart, for which the values.yaml is (partly) generated on-the-fly
- Those charts are not downloaded from a helm repo, but are kept locally and updated in the background
- Those charts can be installed with a complicated shell command (complicated as in: With bigger charts totally unusable)
- The UI and config options for the charts are available in questions.yaml
- However: The UI in SCALE is currently hardcoded and NOT based on said questions.yaml file
- The UI is an early demo version of what is to come,for example: The GPU option is already available in the chart config, but the hardcoded placeholder UI isn't adapted to accept it.
- The hostPath mountpoints are also configured in the chart already...
- Messing with config files and/or manually running helm might work out, but might also break things in the future.
- Options for Ingress are currently also missing, it's only hostNetwork or Nodeport.

So:
Everything you want is already done for the charts, you just need to wait a while for IX to update the UI to reflect the actuall current state.

To give you some context:
I'm working my glorious butt off to get some community Apps prepared for the next release of SCALE (see: https://github.com/truecharts/truecharts).
But even I still run my good old docker-compose stack because it just isn't solid enough at the moment.
 

dalnew

Dabbler
Joined
Dec 9, 2020
Messages
26
@morganL Yes I read through that documentation, but unfortunately it touches mostly on the either complete command line users or complete UI users. I am one of those unfortunate suckers trying to to something in the middle. In an ideal world I would like to setup docker containers via the UI (in some very basic form) but then add all the complex configuration in the generated helm charts/config files. Mostly because the UI, at least in its current state, doesn't provide me with all the options I need and it's just slower trying to map in 10 directories into a docker container. And that's ok, because it's obviously beta. I guess I was just a bit confused by what the UI was doing. In my mind, which to be fair has only a vague understanding of helm charts and kubernetes... I'm working on that :), I would setup a docker container via the UI and it would generate some rudimentary helm chart/kubernetes config file. Then i could go into that and really add in the, eg hack in, the rest of the configuration. However it doesn't seem to actually be doing that as the yaml files generated have none of the config options I set in the UI... which is where I was really getting confused.

@ornias
Great to hear you guys are making progress here. Honestly for a beta it has been, mostly, rock solid!

It also sounds like you are saying SCALE should be doing, at least some of what I want, already. Eg taking the limited UI configuration and dropping a local set of yaml files that represent that config. Let me try to summarize what I've done so far.
1. Setup plex via the default built-in config.
This works if I don't change anything but I start to see weird "crashes" that prevent deployment if I change default config options.

2. Set up plex using the "manual", "Launch docker image" option in the UI.
This works, and I can configure *most* of the things I need, but not all. Some lacking options, Nvidia GPU configuration, setting up directories outside a "pool", etc. When I do this do see a set of generated yaml files, however they don't match at all to anything remotely close to what I configured in the UI. For instance... in my plex config I specified a repo of
"plexinc/pms-docker"
and tag of
"plexpass"

However in the ix-applications/releases/plex/charts/2012.0.0/values.yaml file I see this:
image:
repository: debian
pullPolicy: IfNotPresent
# Overrides the image tag whose default is the chart appVersion.
tag: "latest"

which just looks like it copied the default config in there. The weird thing is the plex app *does* work... to some extent. It at least pulls the right image/tag. So those yaml files have nothing to do with the actual image/container that's running.

3. Set up docker containers via docker-compose
This works almost 100%, with only a couple things I couldn't figure out without "hacking" too much. Those things are:
- How to enable/passthrough nvidia GPUs into the docker via docker-compose without nvidia-docker
- One container seems to have networking issues which don't exist on my other Ubuntu machine (same exact docker-compose.yaml file). It's related to VPN configurations. Normally I expect to see something like:
Code:
Creating TUN device /dev/net/tun
mknod: /dev/net/tun: File exists
Using OpenVPN provider: NORDVPN
Provider NORDVPN has a custom setup script, executing it
2020-12-27 20:12:37 Checking curl installation
2020-12-27 20:12:37 Removing existing configs
2020-12-27 20:12:37 Selecting the best server...
2020-12-27 20:12:37 Searching for group: legacy_p2p
2020-12-27 20:12:37 Searching for technology: openvpn_udp
2020-12-27 20:12:37 Best server : us5848.nordvpn.com
2020-12-27 20:12:37 Downloading config: default.ovpn
2020-12-27 20:12:37 Downloading from: https://downloads.nordcdn.com/configs/files/ovpn_udp/servers/us5848.nordvpn.com.udp.ovpn
2020-12-27 20:12:37 Selecting the best server...
2020-12-27 20:12:37 Searching for country : US (228)
2020-12-27 20:12:37 Searching for group: legacy_p2p
2020-12-27 20:12:37 Searching for technology: openvpn_udp
2020-12-27 20:12:37 Best server : us5848.nordvpn.com
2020-12-27 20:12:37 Downloading config: us5848.nordvpn.com.ovpn
2020-12-27 20:12:37 Downloading from: https://downloads.nordcdn.com/configs/files/ovpn_udp/servers/us5848.nordvpn.com.udp.ovpn
Starting OpenVPN using config us5848.nordvpn.com.ovpn
Modifying /etc/openvpn/nordvpn/us5848.nordvpn.com.ovpn for best behaviour in this container
Setting OpenVPN credentials...
adding route to local network 10.0.0.0/16 via 172.19.0.1 dev eth0

but instead see:

Creating TUN device /dev/net/tun
mknod: /dev/net/tun: File exists
Using OpenVPN provider: NORDVPN
Provider NORDVPN has a custom setup script, executing it
2020-12-26 21:41:58 Checking curl installation
2020-12-26 21:41:58 Removing existing configs
2020-12-26 21:41:58 Selecting the best server...
2020-12-26 21:41:58 Searching for technology: openvpn_udp
2020-12-26 21:41:58 Unable to find a server with the specified parameters, using any recommended server
2020-12-26 21:41:58 Best server :
2020-12-26 21:41:58 Downloading config: default.ovpn
2020-12-26 21:41:58 Downloading from: https://downloads.nordcdn.com/configs/files/ovpn_udp/servers/.udp.ovpn
curl: (6) Could not resolve host: downloads.nordcdn.com
No VPN configuration provided. Using default.
Modifying /etc/openvpn/nordvpn/default.ovpn for best behaviour in this container
sed: /etc/openvpn/nordvpn/default.ovpn: No such file or directory
sed: /etc/openvpn/nordvpn/default.ovpn: No such file or directory
sed: /etc/openvpn/nordvpn/default.ovpn: No such file or directory
Setting OpenVPN credentials...
adding route to local network 10.0.0.0/16 via 172.17.0.1 dev eth0
Sat Dec 26 21:42:18 2020 disabling NCP mode (--ncp-disable) because not in P2MP client or server mode
Options error: You must define TUN/TAP device (--dev)


It seems to be DNS related, but I'm not exactly sure where it's going wrong since my other containers seem to be ok.
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
@morganL Yes I read through that documentation, but unfortunately it touches mostly on the either complete command line users or complete UI users. I am one of those unfortunate suckers trying to to something in the middle. In an ideal world I would like to setup docker containers via the UI (in some very basic form) but then add all the complex configuration in the generated helm charts/config files. Mostly because the UI, at least in its current state, doesn't provide me with all the options I need and it's just slower trying to map in 10 directories into a docker container.
-Thats not how the Apps system is meant to be working, it's designed for users that one a zero-CLI way of doing thing. If you want CLI, you need to use standard Helm charts, which is perfectly possible.


And that's ok, because it's obviously beta.
- It's NOT a beta, its an alpha, meaning it isn't stable nor complete in any way, shape or form.


I guess I was just a bit confused by what the UI was doing.
Some clear flowcharts might be usefull when Beta hits.

In my mind, which to be fair has only a vague understanding of helm charts and kubernetes...
Tbh, the UI doesn't have much to do with Helm or K8s... it's build ontop of those, but what the UI is doing is pretty much limited to generating Helm values.yaml and thats pretty much it...


I'm working on that :),
there is always something to learn. The whole reason I am currently able to work on building Apps, is the fact i've started learning Helm and K8S is september as soon as there was a timeframe and vague design for SCALE Apps.

I would setup a docker container via the UI and it would generate some rudimentary helm chart/kubernetes config file. Then i could go into that and really add in the, eg hack in, the rest of the configuration.
- Helm files are not a "modify on the fly" kind of thing, running stock. You load a helm chart (with your personallised values.yaml) and after it's loaded the config files are not read anymore, they are turned into k8s objects, within k8s. Ofcoarse you can modifiy the values.yaml and update the k8s objects (which can even be done automatically), but thats not the "default" way of doing things in Helm.

- IX Apps also feature a number of additional failsafes and rolebacks for chart updates, which would be totally incompatible when you go edit all sorts of things.

- If you want to run plain Helm files to edit yourself: Use plain helm.

However it doesn't seem to actually be doing that as the yaml files generated have none of the config options I set in the UI... which is where I was really getting confused.
-i'm personally not sure how things are going to be saved. Did you check all files?

Great to hear you guys are making progress here. Honestly for a beta alpha it has been, mostly, rock solid!
;-)

It also sounds like you are saying SCALE should be doing, at least some of what I want, already. Eg taking the limited UI configuration and dropping a local set of yaml files that represent that config. Let me try to summarize what I've done so far.
- yes, but messing with it is:
a. totally unsupported
b. will fuck things up in the future

If you want to use plain helm charts, just go use plain helm charts. You are overcomplicating a LOT of things. You could even create helm charts from docker-compose yaml, which you already have.


1. Setup plex via the default built-in config.
This works if I don't change anything but I start to see weird "crashes" that prevent deployment if I change default config options.
Are you doing so inside the UI? If so what did you change?
Outside of the UI and redeploying the way you are doing IS NOT supposed to be Possible.
(okey, everything is possible, but you get the point)

2. Set up plex using the "manual", "Launch docker image" option in the UI.
This works, and I can configure *most* of the things I need, but not all.
- Thats very well possible, in that case: use your own Helm chart, because the UI is not meant to cover every edge case AND it's an alpha so not supposed to be featurecomplete.

Nvidia GPU configuration
- Expect it next release, it just missed the previous one by a few weeks.

setting up directories outside a "pool"
Thats not going to be possible, all content should be inside the pool, thats how TrueNAS and FreeNAS manage storage. It's unlikely that main design principles of TrueNAS/FreeNAS are going to change.

How do you think features get added when you refer to them as just etc? ;-)

When I do this do see a set of generated yaml files, however they don't match at all to anything remotely close to what I configured in the UI. For instance... in my plex config I specified a repo of
"plexinc/pms-docker"
and tag of
"plexpass"

However in the ix-applications/releases/plex/charts/2012.0.0/values.yaml file I see this:
image:
repository: debian
pullPolicy: IfNotPresent
# Overrides the image tag whose default is the chart appVersion.
tag: "latest"

which just looks like it copied the default config in there. The weird thing is the plex app *does* work... to some extent. It at least pulls the right image/tag. So those yaml files have nothing to do with the actual image/container that's running.
I should've been clearer: Values.yaml is the file containing some chart defaults. (because thats how Helm is designed, not by design of IX)
IX generates the equivalent of a values.yaml, but the original values.yaml isn't used.

To be fair: this was also not clear for me, Had to have the dev explain it to me.
I also was in the understanding it would actually modify values.yaml

3. Set up docker containers via docker-compose
This works almost 100%, with only a couple things I couldn't figure out without "hacking" too much. Those things are:
- How to enable/passthrough nvidia GPUs into the docker via docker-compose without nvidia-docker
Thats a good one (and a pain) afaik.
Intel works solid though

- One container seems to have networking issues which don't exist on my other Ubuntu machine (same exact docker-compose.yaml file). It's related to VPN configurations. Normally I expect to see something like:
Code:
Creating TUN device /dev/net/tun
...
2020-12-27 20:12:37 Downloading config: us5848.nordvpn.com.ovpn
2020-12-27 20:12:37 Downloading from: https://downloads.nordcdn.com/configs/files/ovpn_udp/servers/us5848.nordvpn.com.udp.ovpn
Starting OpenVPN using config us5848.nordvpn.com.ovpn
Modifying /etc/openvpn/nordvpn/us5848.nordvpn.com.ovpn for best behaviour in this container
Setting OpenVPN credentials...
adding route to local network 10.0.0.0/16 via 172.19.0.1 dev eth0


but instead see:
Code:
Creating TUN device /dev/net/tun
...
2020-12-26 21:41:58 Downloading config: default.ovpn
2020-12-26 21:41:58 Downloading from: https://downloads.nordcdn.com/configs/files/ovpn_udp/servers/.udp.ovpn
curl: (6) Could not resolve host: downloads.nordcdn.com
No VPN configuration provided. Using default.
Modifying /etc/openvpn/nordvpn/default.ovpn for best behaviour in this container
sed: /etc/openvpn/nordvpn/default.ovpn: No such file or directory
sed: /etc/openvpn/nordvpn/default.ovpn: No such file or directory
sed: /etc/openvpn/nordvpn/default.ovpn: No such file or directory
Setting OpenVPN credentials...
adding route to local network 10.0.0.0/16 via 172.17.0.1 dev eth0
Sat Dec 26 21:42:18 2020 disabling NCP mode (--ncp-disable) because not in P2MP client or server mode
Options error: You must define TUN/TAP device (--dev)


It seems to be DNS related,
Thats indeed a DNS problem, try pinging and digging from a running container (preferably this one) and trace it like you would any other VM.
But it is either:
a. A general networking problem
b. A DNS issue

but I'm not exactly sure where it's going wrong since my other containers seem to be ok.
Remember most containers don't actively use DNS a lot...
So it might actually be broken in all...


TLDR:
- If you want to setup via GUI, use Apps
- If you want to setup via CLI use Helm
- If you want to use docker-compose: That might work just fine, but is technically unsupported (and has a few bugs)
 

dalnew

Dabbler
Joined
Dec 9, 2020
Messages
26
@ornias Thanks for all the useful info. Sounds like I have some reading to do on help/kubernetes as that may be where I want to be for a while as the UI stuff matures. I'll continue to experiment and migrate my stuff over. I'll keep an eye out on your github for reference.

Interestingly I got all the stuff working in the container via docker-compose, and now I'm able to run nvidia-smi from within the container. However it turns out that Linux kernel 5.9, which is what 20.12 Angelfish upgraded to, doesn't have official supported NVIDIA drivers yet.

The official drivers currently installed are 450.80.02 which have issues with kernel 5.9. There's a ton of internet griping going on about it and I guess nvidia released the experimental 455 drivers which adds 5.9 support, but getting those installed has been a bit problematic via the experimental debian repos. Lots of zfs boot-pool snapshots and I still haven't managed to successfully get them installed.

They exist in the debian experimental repo but I guess I should wait until they are officially supported by TrueNAS and installed by default. Hopefully this is on IX's radar for the next release in Q1 2021. Or maybe in a nightly?

At least I know why transcoding wasn't working in the container now! :)
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
I suggest opening a jira ticket about it, because this is actually a bug
 

stavros-k

Patron
Joined
Dec 26, 2020
Messages
231
hey @ornias, i can't wait to see your charts available on truenas scale, I can do some testing when you got something ready, I don't consider my self container expert, i know a thing or two on docker, helm will be my first touch, but i can give you some of my findings if any.
i've been using unraid for a long time, but it's time to change, once scale hits RC i'll start migrating. Until then i got a spare pc to do some testing/expirimenting.
Also i hope it's ok, i made an app request also on github link you posted!
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
hey @ornias, i can't wait to see your charts available on truenas scale,
Same here, but I might be a little opinionated ;-)

I can do some testing when you got something ready,
Testing is always welcome ofcoarse...
Currently even the charts that are (mostly) finished, are stuck awaiting IX creating an actually functioning Apps UI for SCALE.
So Sadly we all have to hold our horses for a month or two ;-)

I don't consider my self container expert, i know a thing or two on docker, helm will be my first touch, but i can give you some of my findings if any.

Well, the goal of the Apps, is to make it as non-intrusive as possible... So ideally you won't need any helm knowhow, if you do, I failed somewhere... so Guess thats a layer 8 bug on my side in that case :)

i've been using unraid for a long time, but it's time to change, once scale hits RC i'll start migrating. Until then i got a spare pc to do some testing/expirimenting.
BETA might also be interesting, depending on the features you want to use... I expect most single node features to be pretty flushed out during beta already...

Also i hope it's ok, i made an app request also on github link you posted!
Thanks for the headsup, it seems github didn't send me the memo somehow >.<
 

stavros-k

Patron
Joined
Dec 26, 2020
Messages
231
BETA might also be interesting, depending on the features you want to use... I expect most single node features to be pretty flushed out during beta already...

Got 2 VM's Home Assistant and Blue Iris (HA with USB passthrough)
A bunch of Dockers (Plex with nvidia pass through)
All dockers except Unifi Controller(is on eth0) is on host or bridge network.
Some Dockers access paths like docker.sock and some logs for monitoring which on my brief testing seems to work on Apps UI right now (just need to type manually the path)

One thing i didn't see neither as upcoming feature or as existing is the support for plugins (like unraid has). Like extra functionality on UI like GPU statistics, and some other plugins that now i'm thinking about them, 99% of them is to fight problems unraid itself has, so might not be needed!

TL DR I'm already checking for updates every day on my test machine and looking for new things added :) Dev Notes have become my home page lol!
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
As suggested by the developers, its not a given that SCALE would allow us to host our own App Catalogs and easily add them from the UI.
Only "Official" and IX-maintained "Community" Catalogs, are (somewhat) confirmed to be available on launch.

Some of us might want to use totally different Apps or different ways of installing Apps. Just like how people currently use the Guides from @Pentaflake, Scripts from @danb35 or Jailman (aka Me) on the BSD version, which all require the CLI.
If this feature is added, this would allow people like us, to create new ways for you guys to experience SCALE App creation, straight from your trusted GUI.

So I would kindly suggest everyone to give the following suggestion an upvote:
 

tallnerd1985

Dabbler
Joined
Dec 1, 2016
Messages
10
As suggested by the developers, its not a given that SCALE would allow us to host our own App Catalogs and easily add them from the UI.
Only "Official" and IX-maintained "Community" Catalogs, are (somewhat) confirmed to be available on launch.

Since you have been active in feedback on this scene with Scale and container deployment. In regards to the Ingress, Loadbalancers and SSL certs, would I be the only one in thinking a UI version and proper labels in the Chart deployments to work in unison, would make for a dead simple setup in future builds. In the past I've used Traefik setup through my docker-compose script to make everything work together but every once in awhile, something breaks and I have to reread everything along with review some tutorials/forums for what the issue might be so thinking in this sense, I can see a UI part for this might be beneficial in the long run but I could be wrong in this want and being the only one.

And when I say this, I mean reverse-proxying of applications to access via https://plex.exampledomain.com instead of internally
 

ornias

Wizard
Joined
Mar 6, 2020
Messages
1,458
@tallnerd1985
I'm not talking about just my personal projects (which included traefik) here, i'm talking about the option for EVERYONE to add their own custom catalog (not just a custom App, the whole list of self-made Apps) from the GUI.

But to answer your question:
I already have a working prototype of streamlined Traefik Reverse Proxy deployment on SCALE as a SCALE App (without GUI, because:untestable currently) and some other applications. Said design is (relatively) portable to other K8S ingresses (reverse proxies) as well.

Some info about the design specification for (traefik) ingress (reverse proxy) from the UI, I gave in this (unrelated) github issue:

In the future I would love to continue on projects like the above Traefik App, but I also really want to be able to be able to have people help me with development, by just having them add my custom App Catalog, instead of requiring them to add a hack using the GUI or something worse.
Thats why I requested above upvotes on Jira.
 

tallnerd1985

Dabbler
Joined
Dec 1, 2016
Messages
10
@tallnerd1985
I'm not talking about just my personal projects (which included traefik) here, i'm talking about the option for EVERYONE to add their own custom catalog (not just a custom App, the whole list of self-made Apps) from the GUI.

But to answer your question:
I already have a working prototype of streamlined Traefik Reverse Proxy deployment on SCALE as a SCALE App (without GUI, because:untestable currently) and some other applications. Said design is (relatively) portable to other K8S ingresses (reverse proxies) as well.

Some info about the design specification for (traefik) ingress (reverse proxy) from the UI, I gave in this (unrelated) github issue:

In the future I would love to continue on projects like the above Traefik App, but I also really want to be able to be able to have people help me with development, by just having them add my custom App Catalog, instead of requiring them to add a hack using the GUI or something worse.
Thats why I requested above upvotes on Jira.

Sweet, it would be awesome for the custom app UI registry/repo along with finally a streamlined UI for reverse proxy
 

StanAccy

Dabbler
Joined
Apr 23, 2021
Messages
20
Some general feedback on this
  1. some way to improt a docker-compose file would be awesome - this would allow me to port a variety of existing stacks/containers straight onto TrueNAS.
  2. Its not clear to me which docker volume mount commands/instructions map to the TrueNas UI - I see host path volumes, and a separate volumes. Ive tried both, but if the non host path volume are kube volumes, it would be helpful to be able to at least view those volumes in the UI - just a list of them, sizes, etc (what pods are using them).
 

StanAccy

Dabbler
Joined
Apr 23, 2021
Messages
20
Oh, and the Plex config needs a tweak - the timezone selection is not ordered by anything that makes sense to me - its not alphabetical, its not searchable, thus its a royal PITA to configure the right timezone :smile:
 

morganL

Captain Morgan
Administrator
Moderator
iXsystems
Joined
Mar 10, 2018
Messages
2,691
Some general feedback on this
  1. some way to improt a docker-compose file would be awesome - this would allow me to port a variety of existing stacks/containers straight onto TrueNAS.
  2. Its not clear to me which docker volume mount commands/instructions map to the TrueNas UI - I see host path volumes, and a separate volumes. Ive tried both, but if the non host path volume are kube volumes, it would be helpful to be able to at least view those volumes in the UI - just a list of them, sizes, etc (what pods are using them).

I assume you are using 21.04?

For Docker Compose to Kubernetes, the tool is Kompose. https://kubernetes.io/docs/tasks/configure-pod-container/translate-compose-kubernetes/

Its good that Kompose gets tested out by the community and we see what the main issues are.
 

StanAccy

Dabbler
Joined
Apr 23, 2021
Messages
20
I assume you are using 21.04?

For Docker Compose to Kubernetes, the tool is Kompose. https://kubernetes.io/docs/tasks/configure-pod-container/translate-compose-kubernetes/

Its good that Kompose gets tested out by the community and we see what the main issues are.
I tried to use kompose with a couple of docker-compose files, but couldnt get anywhere due to some apparent parsing error on the line 1 of each file. I gave up at that point. Not a great start. Is there no other option?
 

StanAccy

Dabbler
Joined
Apr 23, 2021
Messages
20
I finally got kompose to run locally, but I have no clue how to get this kompose output ( a ton of separate yaml files ) into TrueNas Scale so I can run the two containers defined in the much simpler single docker-compose file. Is this really the best way of supporting this?

If I could have an import directly in TrueNas Scale that took my docker-compose file, created whatever it needed to create, created volumes referenced in the docker-compose, it would be infinitely more useable. The alternative is to have to wait for someone to figure out how to create a helm chart (?) for the applications I want to run, and then figure out how to get those applications into TrueNas - unless they somehow get those applications into the predefined repos that are installed with the platform.

How do you develop / test a new chart? Do I have to have a registered repo (in whatever structure that is) and then push my helm chart data into that repo in order to get TrueNas to be able to try and deploy it?
 

StanAccy

Dabbler
Joined
Apr 23, 2021
Messages
20
When adding docker containers manually, it would be really helpful if I could import a .env file from my local disk for configuring the container (or even load a docker-compose file and then let me select which container Id like to configure) then have TrueNas ask for the .env file referenced, and configure all the ports and directory mounts.
 
Top