proxmox new feature

guletz

guletz

Distinguished Member
Apr 19, 2017
1,651
300
153
Brasov, Romania
Hello to all,

I make a kindly suggestion, and I am willing to help if devs are want to spare some of their precious time.
The ideea is to add monit into the proxmox. I am not an monit expert but I think it could be usefull in situations like this:

- your node X is down, so we move and run all the VM from this node on node Y (without any nfs/gluster/ cluster fs)
- on your node X your VM z is out of space, so we extend your virtual disk with 10 %
- and I can imagine many others test case.

In short words monit is like an automated pilot who can take care about anything your imagination can. I am not a programmer guy, but I can say you could do anything you want with monit. Soon or later things are broken, but monit can fix a lot of thngs.

Or maybe users Proxmox users can do something. .. I see a lot of smart guy on this Forum (not my case ). Nobody can be sure that this is a good ideea (me excluded because monit save me in many bad situations), but we can try !

Any comments are welcome, including the negative. .. mybe are others good tools!?


Post Scriptum: this is a small tutorial about monit (is not the best one, but I try to show a minimal )

http://xmodulo.com/server-monitoring-system-monit.html
 
Last edited:
  • Like
Reactions: Pablo Alcaraz
Monit is great but I do not see the point in integrating it into PVE itself. Monitoring and possibly automated repairs should be "hosted" always outside of the environment you monitor unless you want your nodes and your monitoring down.
 
Hello LnxBil,

Thx for your opinion. We have a minimum cluster (3 nodes). Monit will run on each node. Any monit instance will watch for his own node and for other 2 node. If one node will go down, the rest of monit instance could send alerts, or even more (try to send a wake on lan magic pkg) Or maybe we can restore all VM that was run on the dead node on other online node(we have replication now in 5.x version)

Have a nice day!
 
gsupp said:
As far as restarting VMs that were running on a node that has failed, Proxmox already has a High Availability feature:
Click to expand...

True. But this is available if and only if you have a cluster fs like nfs/others. Some proxmox users do not have this ... And if your cluster fs is broken ... what you can do ?


But if your node is up and running, and also the VM, but this info is not sufficient ? The VM is out of disk space, or .... only imagine any other situations. ANY HA sistem will/can help for this?


Nagios is a good tool, but it not have action (if fail then do this). Monit can do action( if fail then do / else if ... etc).
Like I said, I am not an expert, but good sense and my own experience tell my that several monitoring sistem are best. One single sistem can fail, but not all.
Many monitorig system can only observe a situation, and this is all. It is ok, if you are happy with this. But I think it is more useful to have a self-fix/repair/auto-healing system for proxmox.

I want to go in vacancy , and nobody will need to call me: the VM is down, because nobody was able to delete some snapshots.
 
Last edited:
Thx to all good guys how encourage me regarding thid ideea. For this reason I will try to go forward. Starting with the next month (I will be in vacancy ) I will try to do something (bad or good, it is hard to say now).
I want to monit-rize any core service of proxmox as I can. Then I have something I will post here. Wish me to have good luck ;) The same for any proxmox user!
 
guletz said:
but it not have action (if fail then do this).
Click to expand...

This is not true, please have a look at event handlers which does exactly what you're looking for.

guletz said:
True. But this is available if and only if you have a cluster fs like nfs/others. Some proxmox users do not have this ... And if your cluster fs is broken ... what you can do ?
Click to expand...

If you need a HA cluster, a real one, then there will not be a single point of failure. So your cluster FS is also highly available. What is the point in running a PVE HA cluster without a HA shared storage?
 
You must log in or register to reply here.
Top Bottom