Here are some improvements ideas

i think about something like an alias. like 'dmz2 costumer xxx' instead 'vmbr142'.
 
near 1 day im migration from ESXI to proxmox
this is my idea:
add traffic shaping
and
graph of:
memory usage each vm
cpu usage ech vm
network usage per each vm
 
zitanix said:
near 1 day im migration from ESXI to proxmox
this is my idea:
add traffic shaping
and
graph of:
memory usage each vm
cpu usage ech vm
network usage per each vm
Click to expand...


yes, this is already on our todo list .
 
tom said:
yes, this is already on our todo list .
Click to expand...
sorry for multiple post:rolleyes:
i forget that:
if you can provide Disk I/O controler for taking control some High usage vm, its very good for perventing overload hard disk...:p
 
Last edited by a moderator:
tom said:
thanks for sharing this. (just to mention, quite some parts are already done and released in the current stable.)
Click to expand...

and some big wish:
editing of network settings / bringdes without reboot of the server

it´s not usable for me in production if i have to reboot the hole server and all vm´s wenn adding a vlan
 
svenr said:
it´s not usable for me in production if i have to reboot the hole server and all vm´s wenn adding a vlan
Click to expand...

The idea is to use a static network configuration. Why do you need to add additional vlans while the server is in production use?
 
dietmar said:
The idea is to use a static network configuration. Why do you need to add additional vlans while the server is in production use?
Click to expand...

we are using vlan´s for separating costumers (eg own dmz or vpn) and connect them to external routers or firewalls so nearly each costumer has an own vlan

or for fast building of test/development-environments it would be nice changing these settings "on-the-fly"
 
Checkbox for each VE "Enable Iptables" that sets the following (or similar) in config:


Code: 
IPTABLES="iptable_filter iptable_mangle ipt_limit ipt_multiport ipt_tos ipt_TOS ipt_REJECT ipt_TCPMSS ipt_tcpmss ipt_ttl ipt_LOG ipt_length ip_conntrack ip_conntrack_ftp ip_conntrack_irc ipt_conntrack ipt_state ipt_helper iptable_nat  ip_nat_ftp ip_nat_irc ipt_REDIRECT xt_mac ipt_owner"
 
dietmar said:
Also, those numbers reminds me - here, one perl process need about 120MB, so you could explain above difference that apache starts 2 processes by default (when you use mod_perl), whereas nginx X only starts one as default (because it uses fastcgi)?
Click to expand...

Why not just replace apache + mod_perl with a *very* lightweight webserver that only spawns CGI processes. That way you'd have <10MB for the web server and the Proxmox Perl web scripts would not consume any memory unless they were actually being used (i.e. somebody logs into PVE).

How feasible would it be to make all the mod_perl / embperl scripts just plain old traditional Perl CGI (NOT Fastcgi) scripts? Doing that would save several hundred MB's of memory use when nobody is using the Proxmox interface (which is the majority of the time).

As an aside, a button to launch the VNC process for a VE would be nice so you could connect to the VE's display using a client-side VNC viewer (instead of having to use the included Java viewer).
 
Last edited:
The System Status page in cluster nodes should show free space in the /var/lib/vz partition as well as the root partition
 
Brilian piece of sw with ease of use.

My wish list for next version:

a clone button

A number of backups to keep
 
Last edited:
The main thing I missed is the lack of support for disk encryption during the installation of Proxmox.

Ubuntu is making it really easy, maybe something to base from (they have a "Configure LVM with encryption" option that does everything automagically and only asks for a Passphrase).

The downside is that on reboot, the server will stop the boot up process to ask for the passphrase. On the other hand, you only have to type the passphrase in one place to have all VMs 'unlocked'. The alternative is to configure encryption on each VM which is a lot more work and the VMs themselves also will stop the boot-up process, but now the admin has to type the passphrase in all VMs with encryption.

For my needs, I find the solution of encrypting the filesystem at hypervisor level to work best.
 
I have a good suggestion that would be fantastic.

Some kind of IP management system, with DHCPD support for automating provisioning for platforms :)
 
DHCPD would be great, but I wonder if it would conflict with one of my VMs that already acts as DHCP and DNS for my network. I'm no expert in such a thing tho.

While on the topic of suggested improvements, could we add:

- some kind of hard drive temperature reading pass-thru from the host to the VMs? not sure how this is achieved
- real DVD-ROM pass-thru from host to a VM that I'd like to exclusively control the drive while it's up.
- improvements to the bridging/bonding... perhaps simplified in the GUI even more as bonding my 3 interfaces has been a headache (I love the VMWare implementation of interface bonding and bridging)
- real cloning function for VMs in the web GUI
- BIG ONE: update the VNC/console implementation to allow for tab and other standard keyboard input to work.. going to a menu for every key is very inconvenient

... Otherwise, loving my Proxmox machine now that my motherboard and memory is all fixed up (it's amazing what bad hardware will do to a hypervisor!)
 
sgtfoo said:
- BIG ONE: update the VNC/console implementation to allow for tab and other standard keyboard input to work.. going to a menu for every key is very inconvenient
Click to expand...

"tab" works on all java versions now, fixes were uploaded today.
 
You must log in or register to reply here.
Top Bottom