FreeNAS 11.2 new GUI suggestions and discussion thread

[nojohnny101]

I've been following this discussion and there as been a lot of good submissions and feedback, I know the developers appreciate it.

As I read @BloodyIron's idea, I agreed with it. I can see where @danb35 is coming from as well. The only real benefit I can personally think of (obviously in addition to @Chris Moore's) is for the purposes of troubleshooting. If you look at any thread that deals with drives, usually anyone assisting others asks for smart data for the offending drive. While this is trivial to query from the CLI, it not so much for most people. In guiding newer users or others who are unfamiliar/reluctant to dive into the CLI, it would be a whole lot easier to tell them to "click this button, click this button, click that button, then copy and past the results". There is is even an excellent resource written by @joeschmuck dedicated to helping users read and interpret the SMART data of drives, which indicates its frequency in troubleshooting drives.

How much of a priority should this feature care? IMHO, maybe under "nice to have". There are obviously numerous other higher priority things for the team to tackle.

Anyways, keep the ideas flowing, looking forward to 11.1 and beyond!

 

[Ericloewe]

Accessible SMART data would be useful. A very simple monospace font window would really be enough.

 

[Greg_E]

Not sure if this was mentioned yet... On my 11.1RC install this new GUI is super slow compared to the traditional interface. I'm running it on a real hardware server so this is not a VM issue. Not sure why it is so much slower, but it is definitely something I noticed. Look and feel wise, it's different. Things change, progress progresses, etc., the new look isn't something I'm going to get bent out of shape over and I'll adapt or die (as the phrase goes).

Server specs. if it matters:

SuperMicro Super Server with x10 mainboard
32GB of RAM
dual Xeon E5-2306v4
slow USB boot drive (in need of an upgrade to faster USB3 drive)
256GB Samsung SATA 850 Pro SSD for SLOG and L2ARC
eight 1TB HGST spinning drives for the array
dual onboard 1GB NIC with LACP for failover
Enerasys (Extreme Networks) C5 series switches in a stack - LACP ports spread between two switches

This server is not fully configured yet, still running as a DHCP client since I'm still in testing. Needs to be production ready in January to replace a Windows file share.

 

[Ericloewe]

What browser?

 

[Greg_E]

I was using Firefox, I can try Chrome next time I have the time to fiddle with it. The difference in speed was glaring.

Now all that said, this was being rendered by a little Atom dual core processor. This might be a case where the more simple style of the old UI worked in favor with this processor. I'll probably bring in a more powerful laptop to finish setting this up, that would give me a bit of insight into what is slowing things down.

I'll get back to this pretty soon and concentrate on it, it's almost time to have this online for our students next semester, with the holidays between here and classes there isn't a lot of time for me to get everything set up the way I want to take it. Thankfully I can always use the classic UI to get the work done.

 

[Ericloewe]

It could very well be that the UI is giving the browser a hard time, for some reason. See if the developer tools give any useful indication.

 

[Stilez]

Just wanted to chime in for a quick second. While I've been quiet on the forum, I've been keeping an eye on this thread from time to time. What I can tell you right now is that we are aware of the problems you guys have mentioned so far. They are all on our radar and we are working through our to do list as quickly as possible. This includes but is not limited to:

SNIP

Our plan is to first get the UI to a relative feature parity with the old UI. The upcoming 11.1 strives to get us as close to that objective as possible. For 11.2, we will be working hard to get the UI to a state where it will be default worthy. Most of the refinement and fine tuning will happen for the 11.2 release. Anything before 11.2 is a technology preview so please keep that in mind.

I want to thank you guys for taking the time to give some constructive feedback. It was comforting to read through it and find that the team has also noticed a lot these same issues. It let's us know we're on the right track. This is a huge undertaking and we're working hard to make it as great as we can, since we on the UI team are also FreeNAS users. Thanks for being patient while we work towards our goals. Speaking of which, I need to get back to it. The 11.1 release is coming up soon.

I just wanted to say that this kind of post is worth its weight in gold. It lets us out here know where things are. Sometimes, especially with big project changes, it can happen that the user base feels unheard or isn't heard, or someone thinks they "know what's best for everyone" or that simple (or whatever) is *it*, and doesn't listen. (Even if the intentions are good). So thank you for the detailed comprehensive and reassuring post

 

[Stilez]

One comment on the SMART thing. A useful feature would be a graph of smart data over time, for selected data. For example, it might be that the disk is healthy, but still some data is a bit anomalous. I might want to know that, track it, or keep an eye on it - and I will find that easier if it's in the context of a graph showing trend over time, not just "todays number".

I also know plenty of people (I'm one, when it comes to SMART) for whom a one-word "Healthy" just doesn't cut it. Because healthy by whose criteria - the daemon author? the middleware author? (As a crude analogy, imagine "volume status" just said "no space issues" rather than how many TB was used and total capacity. It's similar logic, but would anyone be really happy with that information?) Maybe someone wants to be a little more informed. We like to see the actual data the decision was made, upon. Then we'll *know* if it's healthy, and be reassured.

A small point, in reply to the "why do you need more data" question

 

[gerry_the_hat]

Maybe someone wants to be a little more informed. We like to see the actual data the decision was made, upon. Then we'll *know* if it's healthy, and be reassured.

I absolutely agree. As a principle, show more info if unsure and let the user decide what to do with it. I mean, graphing and trending is a key thing.
I wish the disk drives temp would also be graphed, and I think there's a case open for that already.

What I don't like is the argumentation one finds when discussing whether this or that graph would be useful. Again, show as much as possible and let the user decide. I mean, most of the FreeNAS user are prof / semi-prof user, so they know what they're doing, one could think.

Prioritising things and resources is another story. A heard user feels better, than an unheard.

Cheers

 

[Stilez]

I absolutely agree. As a principle, show more info if unsure and let the user decide what to do with it. I mean, graphing and thrending is a key thing.

That might be a very good philosophy, if it is practical. Allow a wide range of graphing and charting - and let the user create custom table and graph report layouts, based on their own wishes and needs. This might not be so hard - one would need a way for sources of useful data to be readable in a standard table/chart producing form, so that anything from arc stats to uncorrectable sector errors can be used as input data. Then the user who wants to customise a chart or graph would need to choose some data (and how to scale each in a graph, or abbreviate each in a table), and the update frequency, and at that point it's almost identical to what we have now.

If you know Microsoft's performance monitor (perfmon.exe on any Windows platform), or the old pfSense system graphs, it's very similar. Metrics are grouped by type or source, the user chooses a metric type (network, disk, zfs, ...) and the exact metric from that group (collisions in/collisions out for network, cache hits/cache misses for ARC) and which devices to report (and/or total for all devices). The layout can be saved for reuse. The graphing code does almost everything else - which it already does anyway in the current UI so the only difference is input choices.

 

[gerry_the_hat]

That might be a very good philosophy, if it is practical. Allow a wide range of graphing and charting - and let the user create custom table and graph report layouts, based on their own wishes and needs. This might not be so hard - one would need a way for sources of useful data to be readable in a standard table/chart producing form, so that anything from arc stats to uncorrectable sector errors can be used as input data. Then the user who wants to customise a chart or graph would need to choose some data (and how to scale each in a graph, or abbreviate each in a table), and the update frequency, and at that point it's almost identical to what we have now.

Wow, thought about that too, but decided not to "overdo" things here, as this sounds like a real big story. I'd like to suggest that the devs make a note and keep this in mind when doing some kind of strategic planning, you know. Let's first get the UI polished in order to abandom the old one, right? See, there're so many open bugs and feature requests, they could code 24 hours a day...

 

[danb35]

Maybe someone wants to be a little more informed. We like to see the actual data the decision was made, upon. Then we'll *know* if it's healthy, and be reassured.

Specifically with respect to the SMART data, the system alerts you (and emails you, if you have email set up--which you should) if any of the SMART parameters show issues. It does this whether or not you have tests scheduled, as long as the SMART service is enabled in the first place. It shows an alert for pending sectors. It shows an alert for offline uncorrectable sectors. It shows an alert for multi zone errors. It shows an alert for internal disk errors. If any of these parameters is non-zero, you get a big red flashing warning light in the GUI, and an email to the root user. If the light is green, then either (1) SMART isn't enabled (in which case there really wouldn't be any data to show with your, or @BloodyIron's, proposal either), (2) all those parameters are zero (as they should be for a healthy drive), or (3) you've already silenced a previous alert for one of those parameters on one of your disks, and it hasn't gotten worse since then.

So, on the third point: If da1 shows one pending sector, you'll get the GUI warning (and an email). If you click on the warning light in the GUI, you can uncheck that warning, and the light will go back to its nice calming green. If da1 then picks up a second pending sector, you're back to the flashing red light in the GUI and another email.

So you want to see the data. The baseline for the data is zero; if any of it is other than zero the system will tell you (and you'll see the data--it will tell you which parameter, on which disk, is problematic, and exactly what its current reading is). So what does your suggestion add?

Now, a couple of points of clarification: (1) I don't think this is a bad thing as such--if it can be implemented without adding excess clutter to the GUI, I don't think it would hurt anything. I just don't see that there's any real value to it, and I haven't seen much attempt at an argument to the contrary. (2) I'm basing my discussion of the alerts on what the current (old) GUI does. I have FN11.1-RC running on a test box, but it doesn't have any disk errors for me to see how it alerts on those. So I'm assuming the new GUI will be similar in this regard.

 

[Brezlord]

I think it would be a good idea to have the host name of the system across the top to clearly identify the system you working on. This would be helpful if you have more than one FreeNAS system to manage. See attached picture.

 

[danb35]

I think it would be a good idea to have the host name of the system across the top to clearly identify the system you working on.

Agreed. It wouldn't even take up any screen real estate.

 

[Chris Moore]

I think it would be a good idea to have the host name of the system across the top to clearly identify the system you working on. This would be helpful if you have more than one FreeNAS system to manage. See attached picture.
View attachment 21518

I totally support this although the font could be smaller, but I do have multiple FreeNAS systems to administer and that has been a pain for me.

 

[Stilez]

I think it would be a good idea to have the host name of the system across the top to clearly identify the system you working on. This would be helpful if you have more than one FreeNAS system to manage.

At the risk of a "me too" post - me too!

 

[tenjuna]

@danb35 I think we all hear where you are coming from, cluttered GUI = not ideal. I agree with you, but I would also like to see at least a summary view of SMART data for all of my disks in one place. For systems with only a couple of drives the CLI is not a big deal, but in systems like mine with 12/24/48 drives it becomes tedious trying to find specific data. Impossible? No. I can deal with it as-is in the CLI, but it would be incredibly helpful to see SMART data in one place. I like to keep an eye on data this isn't necessarily alerted for, such as read error counts, drive hours, that kind of thing. I especially liked the earlier suggestion about having information graphed, as it can be a very easy way to pre-empt issues before you even get alerted for it.

Not a need, to be sure, but a very nice want.

 

[mow4cash]

How close is the new UI to be considered fully usable? I don't want to keep adding input if it's not needed and still a work in progress. I noticed the nightlies made the new UI the default but thought it has a long way to go before being a replacement. I wish there was more transparency with what is going on besides having to read through a ton of tickets but I understand they are busy working on the tickets.

 

[Ericloewe]

It's hard to quantify, but it's supposed to be "ready" for 11.2 - meaning fully usable, minus some obscure detail that got overlooked, I imagine.

 

[majerus]

In the vm card view I think something like this would be a nice touch of polish. Thinking though this you would just be asked when you created the virtual machine a "type", Windows, Linux, FreeBSD following that clarification of specific version if it matters for the image to be called.

Overall very nice work on the whole UI!