Problem/Justification
Since joining this forum, I witnessed multiple users losing their pool, because they went about the wrong way, when they virtualized Truenas.
Impact
Users will less likely lose their data.
User Story
When installing Truenas, a big fat warning message will appear, telling the user that virtualizing Truenas for non experimental purposes is not recommended. It will also give instructions (e.g. passthrough of a controller, not single discs, possible blacklisting of the HBA) of the “proper” way if virtualization is needed.
There’s still a resource on this on the old forum. Big warnings in TrueNAS come too late because one should really look this up in advance and design the system accordingly.
This is true, but sometimes people who want to do the best don’t understand the technology but if they got a pointer would follow those instructions.
The easier you make technology to implement, the more you will get users who don’t understand the technology and don’t understand the details, and the more you need to do to prevent them from doing something stupid.
Even then, history shows that the most intelligent and educated and technical people can still make mistakes, like in 1999 rocket engineers causing a Mars lander to burn up and crash due to confusing metric and imperial units of distance and weight.
And when we see multiple posts here (or on Reddit) about people using SMR disks or failing to virtualise properly or adding a vDev as the wrong irremovable type, that is IMO a clear indicator that more warnings are needed - and not only for the naĂŻve non-technical user because even experienced technically knowledgeable users can make mistakes that they immediately (but too late) realise were wrong.
IMO–and I’ll admit it’s cynical–the people who need these warnings are exactly the people who won’t read or heed them. Conversely, the people who will read, understand, and heed a warning like this will have already done their research and will know what they need.
I was about to add a rant here about iX’ official guidance on the subject, but it looks like it’s improved:
At least it has you pass through the HBA–though it doesn’t talk about blacklisting it for the host OS.
Fester’s Guide is now updated to point to both this and jgreco’s resource on the old forum.
To be fair, users should educate themselves before using a piece of technology: your microwave oven does not remind you that you should not put metal inside it…
It takes a really small effort for a new user to understand which hardware is to be avoided: if people are not willing to do even a simple Google research, that’s their issue.
Unfortunately when a “naive” user fails to do research and then has a problem they naturally come here for help.
We then have a choice - be nice and try to help them or give them tough love and tell them that they are on their own - and because we are nice we use our own time to try to help them.
And thus we are then pragmatically motivated to want there to be safeguards against making stupid mistakes even if we do think that users shouldn’t be so “naive”.
P.S. The microwave analogy would be valid if putting metal in the microwave resulted in an explosion - and the question would then be whether we expect microwave manufacturers to prevent that from happening by 1) including a built-in metal detector that would prevent this from happening or 2) a recorded voice asking them to confirm that they hadn’t put anything metal in the microwave. Yes - such analogies often get very very silly - but I think this is a more realistic comparison.
Alternatively lets use the car driving analogy - would we expect a user to have to take TrueNAS lessons and pass a government regulated TrueNAS test before they were allowed to own and operate a TrueNAS server?
P.P.S. You can probably tell that my use of “naive” is as an alternative to the word I am actually thinking which might just possibly be somewhat less polite.
I’d agree if we were taking about barracuda “value” and purple “security” drives, but unfortunately we’re in the situation where a major manufacturer labeled drives as “suitable for NAS use” which are not.
Searching “hardware for TrueNAS” the first result is the CORE Hardware Guide | TrueNAS Documentation Hub, which is a great starting guide but sadly lacks any warning about SMR drives. This should change.
@Protopia I agree 100% that the user who comes here asking for help, most of the time recognizing his errors, has to be supported and not bashed: this community can give you a TON, which is the reason I stayed.
@Arwen we should have something similar around, albeit a bit different. I see the value of having it in a single, step by step guide for newbies.
At work I get a bunch of Linux SysAdmins, with little to no experience with other software, (Solaris, AIX, appliances), or hardware, (IPMI, SPARC, POWER, HMC, Cisco UCS, IBM DataPower, remote console consolidation devices, etc…). To be fair, most of our Unix computers are Linux…
I ended up writing some training guides just so I can point new hires to them. If they have not read them, and ask a question that is easily covered in one, then I point them back to it. Its not meant to be a punishment. But, with more than a dozen new hires over the last few years, giving personal training is time consuming.
One of the training guides was specifically for Solaris ZFS. Like how to free up space in the root pool. Simply erasing a huge, un-needed file may not free up space if it is being held by a snapshot because of alternate boot environments. The huge, un-needed file still needs to be removed… and will eventually free up space as older BEs get trimmed back.
Some of the concepts can be hard to grasp. I once worked with an experienced Unix SysAdmin that did not understand ZFS snapshots were not like VMWare snapshots. Nor Linux LVM snapshots. While the result is similar, both VMWare and LVM snapshots act like write transaction logs. Totally different implementation than ZFS snapshots. In VMWare & LVM snapshots, removing them takes time to copy all the pending writes back to the main data store. Could take hours, even days if too many changes were made.
Not sure if or when I might get a round TUIT for the ZFS introduction for TrueNAS users, so feel free to write your own if so inclined.
The main reason I don’t publish similar docs anymore lies in my lack of familiarity with SCALE.
Given the trend imposed by iX, I don’t think that a newbies introduction to CORE would be of much use… and I have little time available compared to a year ago.
Besides, we have plenty of excellent guides on the old forum (and some on the new, with the horrible porting that happened). Maybe a tad old, but still useful to the CORE!
There’s a soft statement about SMR in the CORE Hardware guide… but I’d agree it can be clear. Making those requests in docs feedback is worthwhile.
Google favors old docs with other links to them… we can’t fix that.
