The current default Task Manager available in QubesOS comes from Xfce. Good to have one, but still lacking features (compared to Gnome System Monitor, or Resource Monitor in Windows -not the same budget though-, features like temperature, network, resources by VM, etc).
Has someone already tried to improve Task Manager, maybe by replacing it with some System Monitor like stuff (both are using GTK, but is it easy ?), or should a heavier solution be used (like Munin,Nagios, Zabbix, nmon, etc) ?
Temperature (I don’t understand the behavior of the fans, which are noisy)
I think temperature comes from cpu usage : consolidated cpu usage in xfce task manager should be enough, but it cannot help understand when a process loops and eats a core for a long time, so a cores usage is sometimes nice.
network usage, at least consolidated (and I guess network usage per application per VM would be impossible to have) one.
gpu usage … will be useful when I will configure gpu pass-through.
other cool (standard ? ) stuff, such as disk I/O, to visualize time needed for some tasks for instance (e.g. VM launching, typically) …
Yes, as dom0 is separated from nearly everything, I though about distributed monitoring schemes, but pretty heavy to try
Yes, thx. I knew htop, very good software. But of course to have the network I need to install it in sys-net, where … the cpu / ram consumption will be the ones of sys-net, not the whole Qubes OS system
I know this is tricky, but I’m looking for a way to have it working … like a “normal” laptop … hence maybe monitoring software is the solution ?
These two you can set up via the Sensor plugin (main menu → System Tools → Panel → Items → add item). The CPU temp should be general for the whole system, unless maybe if you pinned dom0 to only some cores. You can also monitor CPU usage in the panel, but only for dom0. When a lot seems to be going on, however, while your dom0 monitor isn’t showing much, you can click on the qubes widget (blue cube in the top panel), which will shows CPU and RAM usage for all running qubes, so heavy CPU usage of some other VM should show up there.
For CPU temperature, you can have it with the Acpitz’s Sensor. (or with the conky below)
For Network, you install htop in the sys-net template. Then, launch it in your sys-net. You can see the traffic by typing F2 → available meters → Network IO. And also see the processes of your VM.
I can offer you to install the conky, which is very well done and which will give you a maximum of details on Dom0 but also on all your VMs.
Yes, this is great, thank you.
With Sensors I can find all the data I had in command line in dom0 with … $sensors …
So I will attach as many Sensors as needed (and btw … I still hadn’t began to customize the Panel … thx).
Hadn’t pined dom0 to a peculiar number of cores, I haven’t gone through that level of management yet.
I also tried System Load Monitor, but is a replicate of xfce-taskmanager (maybe what you meant with CPU usage), so only deals with dom0, and Disk Performance Monitor … well, currently I guess I’m not using disk so much.
Yes, qubes widget is very useful showing all these data for running qubes … when clicking as everything seems to boost (nearly perfect).
Thx a lot
Yes, thank you, just found it completing @Bearillo’s answer (btw the way I don’t know replying “point-to-point” like you, very precise and useful, thank you, I’ll learn). So I’m using a Sensor for the fans for instance … well, currently it lacks some data … but temperature is nice.
Okay, htop in sys-net with Network I/O works, great, thx (a bit of learning with htop too).
Well … conky seems to have everything, impressive … and since I’m testing : I’ll test .
… hence maybe monitoring software is the solution ?
