0

I have a KVM Host with 4 vms running 24/7h. All VMs runs smoothly but at a given time, all 4 VMs stops to ping and after 10 minutes, it continues to run.

I knew that they are not pinging because I have a script that pings them on every 10 minutes in order to warn if they are ok. I don't know if it's just a network problem or if the KVM simply stops to work.

How can I trace or log their current activity using libvirtd/virsh tools in order to find where the bottleneck is?

Watch the CPU/Network graph on GUI console is not an option because the problem occurs between 6 and 7h AM (when we are not on office).

I've alread read similar solutions 593140 and 184816 but I was looking for a libvirt or virsh solution instead of install a monitoring tool on each guest.

Josir
  • 141
  • 6
  • Why are you insisting on using a VM toolchain to diagnose a (potential) networking issue before you know what the nature of that issue is? Is there some reason you suspect that this is the place to start? If so, adding that to your question would help to narrow this -- it's pretty broad as is. – Spooler Mar 22 '19 at 14:30
  • In fact, I don't know if it is even a network problem. Why am I insisting on a VM toolchain? Because I don't want to add a new variable to the equation (in this case, the monitoring tool). I think my question is very clear: is it possible to monitor/log a VM using libvirt tool ? – Josir Mar 22 '19 at 18:40
  • Your question is clear, but your problem is not. Using a monitoring tool, sniffing your network, looking at OS logs, gathering additional vital data during the time of failure, and generally following a pattern of investigation is not "adding variables", it's troubleshooting an issue using whatever tool is appropriate for the job. One does not define how a job should be done by statically defining the tool to be used first. That said, there IS some vital data available via libvirt, but only for the VMs themselves - not what the VM is running. You can find VM logs in `/var/log/libvirt/qemu/`. – Spooler Mar 22 '19 at 19:40

1 Answers1

0

If you are using qemu, please check this link you can use by own way to dig the exact issues https://qemu.weilnetz.de/doc/qemu-doc.html#Commands, and yeah of course logs are good place to find issues on host. Hope it will help.

asktyagi
  • 2,860
  • 2
  • 8
  • 25