Disable server shutdown/restart from the web interface
- Thread starter arnolem
- Start date
I agree with the reporter 'arnolem'.
Btw, I've found fantastich Alan Cooper's book "The innmates are running the asylum", really a "must have" for every programmer
Yes, but you get a message also if you do it for a VM, so is easy to make a mistake. Me, for instance, clicking on the node, Search tab (the first one), I see all the vm. I click on a vm and then the button "shutdown", yes, and I shutdown the server. Would be very easy move Restart/Shutdown/Shell on the left side, near the node name. If not possible would be good use red color.
Btw, I've found fantastich Alan Cooper's book "The innmates are running the asylum", really a "must have" for every programmer
+1
Thank you for this tip. But must think of those who have not yet made the mistake and who will one day.
I would definitely vote for this, because I just did the same damn thing, on a remote machine which I cannot power on my self and cannot be powered on until normal work hours ...
I was under the impression that I was shutting down the container I had in context and in the rush I didn't read the message in the pop warning properly
I was under the impression that I was shutting down the container I had in context and in the rush I didn't read the message in the pop warning properly
Last edited:
That is differently not a bad idea to do so, but it isn't a real fix for that, just a work around. Right now I'm evaluating PVE, and currently I'm trying out different stuff, and management of administrative privilege isn't one of the features I have explored yet.
Also I wasn't even aware at that time (I am now...) that you could literally shut down a whole node from the interface
But I will still vote for this, as an improvement, because I still think it's easy to be confused about the visual contexts when node and guest management are so much alike, UI wise.
I think I worked out what wen't wrong for me:
Maybe a red colored button and a written confirmation on the following dialog/modal/popup thingy so you know that you are literally about to saw off the branch one is sitting on, especially when shutdown can "mean" many things in PVE
Also I wasn't even aware at that time (I am now...) that you could literally shut down a whole node from the interface
But I will still vote for this, as an improvement, because I still think it's easy to be confused about the visual contexts when node and guest management are so much alike, UI wise.
I think I worked out what wen't wrong for me:
- I was deleting a couple of containers, for some reason selected the node instead of the container I was going to shutdown in the "server view" left panel.
- I then clicked on the next container in the result table in the search context of the node, coursing the row to be highlighted.
- Believing that I now had selected a container, I then clicked the shutdown button, and pressed OK on the following popup without reading the message.
- A lot of WTF ?!?!
Maybe a red colored button and a written confirmation on the following dialog/modal/popup thingy so you know that you are literally about to saw off the branch one is sitting on, especially when shutdown can "mean" many things in PVE
Why is this a work around? The whole permission system is designed to do exactly such things.
It's a workaround because the design is flawed when you have full permissions, and to fix this problem you suggest to use less permissions (workaround). In other words, you are not fixing the problem i.e. moving the buttons, but suggest to login in a way that prevents it to use them at all.
If I need to login with full permissions, then I still will have to deal with bad design.
I would suggest a red color for button when a node is selected and a red(eesh.. ) tint background for confirm popup.
so if you get several popups you can be sure you are not at the node level.
I do, however, understand it is not overly easy to do.
so if you get several popups you can be sure you are not at the node level.
I do, however, understand it is not overly easy to do.
mmenaz more or less outlined it, why it's a work around and not a fix. So i'll refer to his reply.
I second this. The process to shutdown the server is almost identical to shutting down a VC. I accidentally checked the node before shutting down a VC and didn't switch back and shutdown the entire server instead (which was in a remote data center so it was 20 minutes before we could get it turned back on). Process is just way too similar to shutting down VMs or VCs. So I second a giant red screen that looks nothing like the shutdown confirmation you get for shutting down a VC.
Easy fix:
just run on each proxmox node:
sed -i 's/restartBtn, shutdownBtn, shellBtn, actionBtn/shellBtn, actionBtn/g' /usr/share/pve-manager/js/pvemanagerlib.jsThank you for posting this. I would make one change in leaving the restart button available. Even if you accidentally hit it, and then blindly click through the affirmation, then at least the system will only reboot.
That said these buttons should be removable via an option. SSH, KVM, and remote management(IPMI) are all options for shutting down or restarting a physical server. If an admin has one or more of these then why not make it possible for them to remove buttons that can be accidentally clicked on?
Agree with OP. ESX had "maintenance mode" or something, and you can't just shutdown the host (or a node), there were extra steps. For those of us who do these repetitive actions all day long, it's way too easy to just click OK on the tiny (nearly illegible) popup. We need mechanisms to protect us from ourselves for such operations. When we want to delete a VM, we have to type the VM number. Why not make me type the name of the node to shutdown/reboot? And color that box differently, so it's unique.