I saw a notice from Joplin desktop app that said something like “you need to update the server side or the sync will stop working”. Despite the sync working perfectly fine, I opted to update.
Went into TrueNAS web UI, nothing to be updated. Hmmm…
Refreshed the app catalog, took forever to complete…deleted the catalog, re-added, didn’t work.
Looked up the internet, apparently I needed to upgrade my TrueNAS, which, I just did a few days ago - but apparently there is this new Bluefin thingie, alright then, update it goes!
Update went through, and most of my apps were broken, containers wouldn’t start due to errors like unable to mount volumes, or for using SMB share service. Looked up the internet…and the solution seemed to be - reinstall.
At this point I was just so fed up. This isn’t even the first or second time that something like this has happened. It just seems that every half year or one year, everything gets broken and needs to be redone.
Years ago I was using ESXi and I would spin up a VM for each app. Very clunky and ineffective solution for sure, and yes things did get broken randomly back then, but never have I felt so frustrated and defeated. Why do things have to be broken every half to one year? Nobody needs this hassle to take away their time and energy
I know the pain… That’s why I prefer as much separation as possible. Physical separation, OS separation…
If there are hardware resources, I personally always try to go in the direction of separation just to avoid unnecessary mess.
Keeping all your eggs in one basket is convenient and efficient, but until something goes wrong and backfires on many other things.
I try to limit the number of eggs per basket as much as possible. Even in my home LAN I have two separate machines that have dns(pihole)… why? In the event of unavailability of No.1 for any reason. It is known that not every service can have its own live copy, but the separation allows you to keep at least some % of the services.
So even if you don’t have the ability to physically separate, you might consider going back to OS virtualization. If you don’t want to keep TrueNAS in a hypervisor, you can always try to run virtualization in TrueNAS and separate yourself on the OS layer, not just containers and apps.
When everything is on one OS layer when there is a problem like yours everything can be affected unfortunately. Sometimes even a simple system update can cause a problem. Imho, it is worth using some separation segment or at least a separate test environment that will make sure that the production machine will not have problems.
Before I make any updates or changes, I first test them in a separate environment, keeping the production machines away.
P.S
Always keep a reasonably current backup before doing anything so reinstalling is less painful.
Can’t help but agree. IME, the out of the box apps are either documented badly (i.e. they will create pain points for you to trip over if you follow them), they’re out of date, or the apps are not documented at all – and that’s assuming they even have any hope of working in the first place (looking at you, OOTB Pihole chart). The truenas forum then points you to use TrueCharts. But that’d still leave you in this spot during an update, and the TrueCharts folks say “that’s not supported” seemingly every other breath. Really feels like a tacked on afterthought that nobody wants to support or improve.
Definitely agree on sanity being running docker in a VM or running it altogether separately (whether virtualizing both the docker host and truenas environment, or different machines altogether). Should’ve just done that to start, would’ve saved myself a couple of weeks of frustration.
I just went through this. Updated TruNAS and broke everything. I am honestly debating going back to Proxmox but idk. I am realizing that most of my data is served by docker programs so I don’t know if I really need a NAS more than I need a decent hypervisor that I can rely on to run docker and like 2 VM’s.
This just reinforces in my mind I need to keep finding ways to do infrastructure Automation. I.e. the solution for this with docker and portaier is yet again far more durable and upgradable especially if they leave it alone in the host os, once it’s setup.
I’ve had no trouble moving data for nextcloud, for example, between a docker contsiner, VM and whatever flavor is in Synology app center.
Zero issues. Even going way off script.
I had the truenas plex app but I wrecked it with a. Update and it was not clear to me what I needed to do to recover, or repair the dataset.
The separation of concerns is handled differently with containerization but the effective ability to silo apps can be achieved with the ‘old school’ method of multiple virtual machines just as well as with containers.
It comes down to two different philosophies about managing your stuff; treat it like cattle, or treat it like a pet?
IME, the out of the box apps are either documented badly (i.e. they will create pain points for you to trip over if you follow them), they’re out of date, or the apps are not documented at all – and that’s assuming they even have any hope of working in the first place (looking at you, OOTB Pihole chart). The truenas forum then points you to use TrueCharts. But that’d still leave you in this spot during an update, and the TrueCharts folks say “that’s not supported” seemingly every other breath. Really feels like a tacked on afterthought that nobody wants to support or improve.