New tool: pmmaint

[tuxis]

Hi,

while managing a lot of clusters, we often run into the issue that customers have thought about where they want to have their VMs running. So, when doing maintenance on a cluster, after a reboot of a node, we need to place the VMs back on the node that they were running on. Also, when you want to reboot a node, it's often quite some work to figure out where we can migrate VMs to without depleting memory of a sibling.

So we created pmmaint. It allows you to create a configuration snapshot of the current configuration and tell nodes to empty themselves. pmmaint will figure out where the VMs should fit, taking into account local-storage (partially, it will not think about space) and CPU-types. When you're done, simply type 'pmmaint restore <snapname>' and all your VMs will be nicely migrated back to where they came from.

Version 1.0 is available as a Debian package. See https://gitlab.tuxis.nl/oss_public/pmmaint for more information. If you have any questions, let us know and feel free to improve the code. I'm not a hardcore developer, so I'm pretty sure there are improvements to be made.

Mark

 

[VictorSTS]

I will take a look at this asap, seems useful!

For the part that deals with where the VM/CT should be running, I usually just add a note to each VM with the name of the node it should be normally running in and follow that while moving them in/out of a host for maintenance of after a hardware failure.

I'm wondering: why is Proxmoxer needed? Wouldn't just pvesh be enough?

 

[dlasher]

Nice work - seems like something that would be great integrated into the PMX GUI - have an "evacuate node" right-click-menu option.

Spacing needs a little help for larger hosts, this is my lab:

                    |Memory (GB)
hostname            | total free  used  | CPU
pmx1                | 125.7482.88 42.86 | AMD Opteron(tm) Processor 6380
pmx3                | 125.7471.39 54.35 | AMD Opteron(tm) Processor 6380
pmx5                | 125.8979.20 46.69 | AMD Opteron(tm) Processor 6380
pmx4                | 125.8871.61 54.26 | AMD Opteron(tm) Processor 6380
pmx0                | 251.58162.1789.41 | AMD Opteron(tm) Processor 6380
pmx2                | 125.8886.89 38.99 | AMD Opteron(tm) Processor 6380

 

[adamb]

Alot of this can be done with groups as is.

 

[tuxis]

Nice work - seems like something that would be great integrated into the PMX GUI - have an "evacuate node" right-click-menu option.

Spacing needs a little help for larger hosts, this is my lab:

                    |Memory (GB)
hostname            | total free  used  | CPU
pmx1                | 125.7482.88 42.86 | AMD Opteron(tm) Processor 6380
pmx3                | 125.7471.39 54.35 | AMD Opteron(tm) Processor 6380
pmx5                | 125.8979.20 46.69 | AMD Opteron(tm) Processor 6380
pmx4                | 125.8871.61 54.26 | AMD Opteron(tm) Processor 6380
pmx0                | 251.58162.1789.41 | AMD Opteron(tm) Processor 6380
pmx2                | 125.8886.89 38.99 | AMD Opteron(tm) Processor 6380

Hi,

The memory should be in GB. So I'm guessing that your locale confuses the print on line 707-711. Can you look what you get if you add

print(nodes[node].get_mem())

to /usr/sbin/pmmaint on line 707, and move all lines below one line lower?