Proxmox Load Average: make it better to read?

[kotakomputer]

When Proxmox Load Average show 150 then I can not determine which VPS caused this high load since all VPS CPU < 90%.
Is this a way to show load of each VPS so I can see where the 150 come from?

NB: currently, I must use enter to each VPS then run #top command 1 by 1

 

[guletz]

Hi,

You coud make a script in all of yours VMs/CTs like this(only the basic ideea):

cat /root/top1.sh

#!/bin/bash
echo $(top -b -n1| head -8 | tail -n 1)


Then you could run this in the PVE server:

pct exec ID /root/top1.sh


Even better is to get the proccess who cpu > your_limit and use monit to to check this and if your your_limit is pass, then instruct monit what to do(stop / restart / what_ever for your VM/CT and/or send a mail)

Good luck!

 

[kotakomputer]

I have use Monit, but I need to know maybe a field of "Load avg" on each VPS list?
NB: I know I can use M/Monit but this is not my concern

 

[guletz]

You can use monit for load avg or whatever you want. also you can modify the command line to get load avg from top output!

 

[kotakomputer]

You can use monit for load avg or whatever you want. also you can modify the command line to get load avg from top output!

The problem is w/o m/monit I must visit each VPS 1 by 1. I hope Proxmox have a new Field calles as Load Avg, look like CPU or RAM used.

 

[guletz]

The problem is w/o m/monit I must visit each VPS 1 by 1

NO. It is automated... you wanly check your mail in the worst case!

 

[kotakomputer]

Why must check mail? That's why I suggest Proxmox add a new field called as "Load Avg" on list of VPS.
I hope you understand my concern is feature suggestion of Proxmox not about how to use monit.

 

[guletz]

ok, I respect your remarks. For me, using a stable proxmox cluster, most of the cluster resurces are watched by monit. For most of the cases, monit solve most of them without my intervention. For the rest of events I get mails.

As a simple example, I do not watch that now my iops is higher as normal because monit check if some backup or replication is in use. But in any other cases monit send me a mail if this happens for at least 10 min.

For sure, I can find this info using pmx graphics. But this aprouch is time consuming.

Good luck / Bafta