Re: [PATCH datacenter-manager 0/4] add resource gauge panels to dashboard/views

[Dominik Csapak <d.csapak@proxmox.com>]

On 3/25/26 12:47 PM, Thomas Lamprecht wrote:
> Am 24.03.26 um 11:25 schrieb Lukas Wagner:
>> On Mon Mar 23, 2026 at 12:03 PM CET, Dominik Csapak wrote:
>>> This uses the new pie charts[0] to add gauge panels for resources to
>>> the dashboards/views. Either combined cpu/memory/storage or
>>> indidivually, for pve/pbs or combined counts.
>>>
>>> for this we have to sum the data up in the backend.
>>>
>>> I also added these to the default dashboard (since it's data from an api
>>> call we already query) but put that in a separate patch so we can easily
>>> decide to not apply that. (not sure if we want to change the default
>>> dashboard)
>>>
>>> Note that the pwt patches [0] have to be applied and the package
>>> has to be bumped first.
>>>
>>> 0: https://lore.proxmox.com/yew-devel/20260320160816.4113364-1-d.csapak@proxmox.com/
>>>
>>> Dominik Csapak (4):
>>>    api: return global cpu/memory/storage statistics
>>>    ui: css: use mask for svg icons
>>>    ui: dashboard: add new gauge panels widget type
>>>    ui: dashboard: add resource gauges to default dashboard
>>>
>>>   lib/pdm-api-types/src/lib.rs      |   2 +-
>>>   lib/pdm-api-types/src/resource.rs |  27 ++++++
>>>   lib/pdm-api-types/src/views.rs    |  15 +++
>>>   server/src/api/resources.rs       |  65 ++++++++++---
>>>   ui/css/pdm.scss                   |  35 +++----
>>>   ui/src/dashboard/gauge_panel.rs   | 156 ++++++++++++++++++++++++++++++
>>>   ui/src/dashboard/mod.rs           |   3 +
>>>   ui/src/dashboard/view.rs          |  19 +++-
>>>   ui/src/dashboard/view/row_view.rs |  43 +++++++-
>>>   9 files changed, 331 insertions(+), 34 deletions(-)
>>>   create mode 100644 ui/src/dashboard/gauge_panel.rs
>>
>> Thanks for these patches!
>>
>> Some thoughts, some of which we already discussed off-list:
>>
>>    - I'd suggest a wider opening angle for the gauges, so that the gauges
>>      are a bit more compact and look less like a horse shoe
>>      To me, 75/285 looked quite good, but this is of course highly
>>      subjective, so no hard feelings
> 
> Didn't bother me that much when initially looking at it, but you might have
> a point, especially w.r.t. how the text below the gauge fits visually.
> As of now the angle of the gauge edges seem slightly "pointy" w.r.t.
> horizontal line the text gives off (for the lack of a better description).
> 
> FWIW, Grafana defaults to then moving the value inside a bit lower, so that
> it basically lines up with the bottom edges of the graph, no hard feelings
> here, just mentioning it as comparison.

i talked off list with lukas yesterday, and I don't have anything,
against using this wider angle. I'm just used to 3/4 gauges probably.

> 
>>    - For the percent-label inside the gauge, I think 0 (zero) significant
>>      digits are okay, for such a 'global' infrastructure gauge two
>>      significant digits are hardly useful, I think
> 
> one might be OK, at least for <1 and >99 %, and then we might just always
> add it. but, tbh, using zero for now might be fine too. For large setups
> a single decimal point in percentage might be still some sizeable real
> absolute amount though.
> 
> A space between value and unit might be nice.

ok, i'll change it to sthg like '23 %'

> 
>>
>>    - We probably should use full product names (Proxmox VE, Proxmox
>>      Backup Server) in the card headers, instead of just "PVE" and "PBS".
>>      The overall text length could become quite long though, so maybe we
>>      need some other approach for the card title.
> 
> +1, but no good solution from top of my head, might be easiest to live
> with the long texts.

yeah i think so too. I'd change it to 'Virtual Environment' and
'Backup Server' like the other widgets. (no need to repeat proxmox here
on every widget IMO)

> 
> Additionally:
> For widgets multiple gauges placing the icon+label inside or below the
> gauge might look a bit better. At the top it looks slightly out of place
> as is. Or written differently: there's now some info on top, in the mid
> and below, so visually many differnet locations a users eye has to wander,
> feels "unruhig".

i get what you mean. I'll think about where this fits better
(shortly though about simply putting it in the status line, but that
makes the text there often too long)

> 
> Small nit: For PVE CPU Usage it's not immediate clear how the load values
> and the allocated relay to each other.

mhmm do you think it'd be better to leave out the 'allocated' part
completely for now?

it's a bit of a mixed representation anyway, since e.g. i missed
adding that info for memory and it only shows for PVE graphs.

otherwise maybe adding a unit/short explanation makes it better?
---
1.5 of 45 threads/cores used (13 allocated to guests)
---

(not sure if threads or cores is better here)

> 
> Font is handled by the SVG, how is the font family and sizing chosen there?
> does the size always scales nicely?
> I'll try to reply on the yew-devel series too, but from a quick check I did
> not found any note w.r.t. this.


font-family and size is inherited from css, where we set a global font 
family/size. i chose the svg sizes so that such percentages should fit
well inside but one can override that from outside with e.g.
'font-size: 10px'

atm, the piechart does not have any configs for text positioning.
i didn't need it yet, and auto positioning/sizing (aside from centering)
is cumbersome in svg.. (e.g. converting from svg to browser coordinates, 
etc)