FreeNAS 11.2 new GUI suggestions and discussion thread

[Damianos]

A console that listed the VMs and their state...

...Out of scope (maybe) but something to think about is a way to join multiple FreeNAS system so they can be managed / monitored from a single web GUI. If I recall, there was some work on that in Corral but I think that may have died with Corral.

Thanks for the feedback! In the future, I'd like to ask anybody giving multiple suggestions to please format it as a numbered list. This will make things more efficient.

1. A console that listed the VMs and their state (up, down) and gave options for editing their condition or creating them to begin with.
2. It would be very nice if the whole process could be done through the GUI and with the ability to snapshot the VM and spawn a new VM from the snapshot.
3. Mount and UN-mount an ISO.
4. Spawn a viewer into another window so the hardware configuration can be seen in one window while the UI of the VM can be seen in another.
5. Multiple UI windows should be possible so that each of the many potential VMs can be seen at the same time.
6. All the virtual hardware should be configurable through the web GUI such as the number of CPU cores, amount of RAM, virtual drives (or multiple drives) number and type of virtual network interfaces, how the virtual network interfaces connect to the world, virtual switch (or multiple virtual switches)
7. Ability to export the virtual machine so it can be moved to another server and an ability to import the virtual machine that was brought from another server.

These are good. Keep them coming!

 

[Ericloewe]

Snapshotting all VMs at once, after they're stopped, sounds like an easy and useful feature. Might even be a good use for a ZFS channel program, but that part would be in the middleware.

 

[Chris Moore]

All the virtual hardware should be configurable through the web GUI such as the number of CPU cores, amount of RAM, virtual drives (or multiple drives) number and type of virtual network interfaces, how the virtual network interfaces connect to the world, virtual switch (or multiple virtual switches)

I am attaching a snip from a diagram to give you an idea what I had in mind with regard to virtual switches and the ability to have multiple network interfaces assigned to a single VM with each virtual NIC connected to a different Virtual switch and each virtual switch either being private (no outside connection) or able to be assigned to a specific physical interface.

The current management of physical interfaces isn't very smooth but what ever changes are made to the settings need to be such that it doesn't disconnect the web UI.

 

[BloodyIron]

My first bit of feedback. I actually like the 11.0-RELEASE GUI more than the 11.1-RC1 version. And yes, I'm talking about the "beta" GUIs in each, not the default GUIs.

The one in 11.1 is definitely better than 9.10 and earlier. But so far in my testing, the 11.0-RELEASE beta GUI works better for me than 11.1-RC1. The 11.1-RC1 seems less refined...

But I'm testing the 11.0-RELEASE one further. With 11.1-RC1, I couldn't get any VM to start :(

 

[brando56894]

I actually like the 11.0-RELEASE GUI more than the 11.1-RC1 version.

What's different?

With 11.1-RC1, I couldn't get any VM to start

Try the nightly from today or yesterday, I finally managed to get an Arch VM to boot up and I could interface with it via noVNC.

 

[Chris Moore]

Something that needs to be in the new UI is an ability to manage SMART tests and Scrubs and such on all installed disks regardless of if they are in the boot pool or other pool. Right now, there is no consistent interface for it. If you are using disks for the boot media, there is no way to run a SMART test on those disks except to setup a Cron Job to run the command. The menu for setting up SMART Tests ignores disks that are part of the boot pool.

 

[BloodyIron]

1. I'm not sure if the GUI libraries changed, but if you put them side by side, you will see the style is very different.

2. It turns out my lab CPUs can't do nested VMs, d'oh! So far as I can tell, this is a hardware limitation (Super old AMD Opterons).

What's different?



Try the nightly from today or yesterday, I finally managed to get an Arch VM to boot up and I could interface with it via noVNC.

 

[BloodyIron]

I agree with this, and would like to add, that I want to see a GUI way to REVIEW SMART test results too! It's rather time intensive to have to kick to the CLI to get SMART results on each disk, and peer through it. By contrast it would be nice to have GUI parsing of this info, I think that could help streamline the review of SMART stuff a lot, especially when you're dealing with many-disk pools!!! :O

Something that needs to be in the new UI is an ability to manage SMART tests and Scrubs and such on all installed disks regardless of if they are in the boot pool or other pool. Right now, there is no consistent interface for it. If you are using disks for the boot media, there is no way to run a SMART test on those disks except to setup a Cron Job to run the command. The menu for setting up SMART Tests ignores disks that are part of the boot pool.

 

[danb35]

I want to see a GUI way to REVIEW SMART test results too!

What more do you need than is already there? If a test fails, it alerts you. Since the tests are pass/fail, I don't know that there'd be anything more to show.

 

[BloodyIron]

Well for starters, seeing whether the SMART tests are actually happening or not. Alerts aren't enough. If the goal is to build a web GUI that enables you to do "everything", then SMART needs to be fully realised in the same web GUI, otherwise we're still kicking back to CLI to read SMART test results, every time... Again, horribly inefficient use of time.

What more do you need than is already there? If a test fails, it alerts you. Since the tests are pass/fail, I don't know that there'd be anything more to show.

 

[danb35]

Well for starters, seeing whether the SMART tests are actually happening or not.

If you scheduled them, they're happening. If the command failed, you get an alert. If FreeNAS can't read the SMART error/status log, you get an alert. If the test failed, you get an alert. The issue with SMART tests is that they don't get scheduled. If, for example, you replaced a disk in your pool, and weren't aware that, despite the GUI looking like it's tying the SMART test schedule to da/ada numbers, it's actually tying that schedule to disk serial numbers (WTF?), you might not have thought to update the test schedule to cover the replacement disk. In that case, you don't have a test scheduled (for that disk), so it doesn't run. But even then, FreeNAS will alert you if one of the parameters goes out of range (e.g., if the disk shows a bad sector).

If the goal is to build a web GUI that enables you to do "everything",

I don't think that is the goal, and I don't think it's an achievable goal in any event. The GUI should let you do all the common administrative tasks. You shouldn't need to regularly go looking for SMART data--the system will monitor it and let you know if there's a problem. Same thing for disk temps (another common request).

Again, horribly inefficient use of time.

...to do something that there's no reason to be routinely doing in the first place.

 

[BloodyIron]

Yeah, I completely disagree. I can't explain it any better than I have, there is a tangible gain to have this, and it's actually not complicated. Why you're resistant to such a change is beyond me, but whatever man. I'm not going to just perpetually argue this with you, I would actually use this.

If you scheduled them, they're happening. If the command failed, you get an alert. If FreeNAS can't read the SMART error/status log, you get an alert. If the test failed, you get an alert. The issue with SMART tests is that they don't get scheduled. If, for example, you replaced a disk in your pool, and weren't aware that, despite the GUI looking like it's tying the SMART test schedule to da/ada numbers, it's actually tying that schedule to disk serial numbers (WTF?), you might not have thought to update the test schedule to cover the replacement disk. In that case, you don't have a test scheduled (for that disk), so it doesn't run. But even then, FreeNAS will alert you if one of the parameters goes out of range (e.g., if the disk shows a bad sector).

I don't think that is the goal, and I don't think it's an achievable goal in any event. The GUI should let you do all the common administrative tasks. You shouldn't need to regularly go looking for SMART data--the system will monitor it and let you know if there's a problem. Same thing for disk temps (another common request).

...to do something that there's no reason to be routinely doing in the first place.

 

[Jailer]

otherwise we're still kicking back to CLI to read SMART test results

BUT, there's nothing to read unless there's a problem and you will get an alert if there is one. I don't have as strong of an opinion as @danb35 does on the matter but I do agree that I don't think it would add much value for the effort.

 

[danb35]

Yeah, I completely disagree.

Obviously.

I can't explain it any better than I have

Pity, as you haven't explained anything yet--you've just said you want it.

Why you're resistant to such a change is beyond me

Because I don't see any significant value in it (and you haven't explained why you do), and I'd prefer the devs spend their time on stuff I do see major value in. Because the more stuff you add to the GUI, the more cluttered the GUI tends to get. I'm sure there are other reasons, but those should suffice for now.

But really, what do you hope to gain with this? To see that the test ran? If you scheduled it, it ran. To see if it passed? If it didn't, the system would alert you. To see if the disks are getting too hot? The system will alert you about that too. To see if your disks are developing bad sectors? Once again, the system will alert you to that. And it will send all those alerts to your email, so you don't even have to have the GUI up to get them.

So, yeah. When I see your suggestion, I'm seeing more clutter in the GUI, and more work for the devs, with no tangible benefit. And I'm seeing your inability, or disinclination, to explain the benefit you see in it, beyond simply saying you want it.

Edit: To be clear, as long as it didn't overly clutter things, I wouldn't mind having this in the GUI--I just don't see any real value to it. But as long as developer resources are finite, I'd prefer those finite resources be used for stuff that seems to have value (even if it isn't stuff I'd personally use).

 

[BloodyIron]

Pity, as you haven't explained anything yet--you've just said you want it.

Yeah this level of hostility is uncalled for dude. I'm not going to respond to you any further as you clearly have no interest in actually reading or considering my position. I've considered yours, and you just don't even care that other people have different needs than you.

 

[danb35]

Yeah this level of hostility is uncalled for dude.

I think you need to recalibrate your hostility meter--it's giving false positives. I'm not being hostile in the least, to you or to your idea. I'm not inclined to agree with it, but I've asked, very explicitly, on two different occasions, for more information about what benefits you perceive to your idea. But rather than answer me, you seem to prefer to play the victim to mean old @danb35's "hostility." If that's what floats your boat, go ahead--you won't hurt my feelings--but it doesn't do much for a productive conversation.

I've carefully read your posts, but I haven't read or considered your position, as you haven't stated it other than "I want it." Specifically, you haven't said why you want it. I don't know what you've read or considered of my position (as I can't read your mind), but you haven't addressed it in the least.

 

[Chris Moore]

I have a fair variety of storage systems that I must care for at work and several of them use customized Linux installations with a web management console. We don't have the luxury of email communication for these systems because, even though they are on networks, the networks are isolated for security reasons, so I have to access the network and navigate to the webpage to check their status. It is all rather cumbersome but those systems, some made by QNAP and one that was made by Netgear, all have a way to set the drives to do SMART tests and review the results of those tests in the web interface. It isn't easy to get to, but it is there and the reason it is important is because you can't always get notified by email when there is a problem. These systems also log when errors happen so you can just check the log, but it still requires going to the system to do that and it is convenient to be able to access the results of the SMART test via the GUI so you can determine the nature of the error when one has occured.
So, a convenience that would only be used occasionally. Not a high priority and then there are the scripts that the community has produced. I have modified some of the scripts that would normally email results to save them to a directory on the shared drive so the results can be reviewed remotely instead of even needing to go to the system web GUI, but wouldn't it be nicer to see all those results in one interface, that is what menu options are for. If you want it, you can drill down through the menus, but the top level can be streamlined.

 

[danb35]

We don't have the luxury of email communication for these systems because, even though they are on networks, the networks are isolated for security reasons,

Now that's something I hadn't considered. In FreeNAS, it would still light the flashing red warning light in the GUI, but yes, that makes a point in favor of exposing more of the SMART data.

 

[diedrichg]

Throwing my 2 cents in:

Stuff others noted, but would like to second:
-Historical data retention
-use of space/config (but i understand that the objective is to get it working first so understandable...)
-Disk details/view. Would like to be able to get info about the actual hardware
-logs
-mouse overs with info/sample data

Don’t think anyone has noted these ones:
-Task outcome: I can schedule tasks, but I wish I could see the result of them as well in the same place (eg last smart test performed on X date, taking Y time, etc).
For init scripts I would like to be able to see if the script is still running (my fan spin script doesn’t show up as a running process, have to login via ipmi KVM to see the log output displayed every few min)
-Mobile view: probably not a high priority at all, but I use my phone regularly to make minor tweaks/see the status of the system. The new UI is the least mobile friendly yet. The new UI is almost useless on it (hard to be specific: stuff isn’t displayed, charts don’t form, things that should be on the screen aren’t, but the system believes they are, etc). The space use also plays a role here
-I like the VM stats option, if something like that were possible for jails that would be cool (but then again I don’t plan on using jails for much longer)
-Multiple/persistent shells: either or, or both, would be nice. I know I should SSH in, but sometimes what I’m doing is really simple and I’d rather just do it via the GUI, but it may require a second screen or console for a jail, etc
Also, the current console is unusable on a phone (can’t engage with the window object)
-Some customization would be nice. EG if we would like a view with only a few bits of information available, but shown in a condescend format that would be awesome. Going a step further, being able to arrange the main menu items as different window objects would be cool as well (I’m probably dreaming though haha).

Thanks!

You need to create a feature request on bugs.freenas.org if you want your voice heard and your ideas to be considered.

 

[Chris Moore]

You need to create a feature request on bugs.freenas.org if you want your voice heard and your ideas to be considered.

While you are correct, if you refer to this post:
https://forums.freenas.org/index.ph...ions-and-discussion-thread.58815/#post-416013
I think you will understand why the discussion in this thread has included some talk of feature requests.