Sunday, November 4, 2018

Packet dropping between switch and server - I'm out of my depth?

Hi all,

Basically I have been getting a lot of alerts from Netdata about packet drops on my server in my homelab.

net_packets.enp0s20f3 CHART inbound packets dropped ratio (was warning for 12 minutes) the ratio of inbound dropped packets vs the total number of received packets of the network interface, during the last 10 minutes ALARM enp0s20f3 FAMILY Recovered from WARNING SEVERITY 

It's a Proxmox 5.2-6 VM host on a Supermicro A1SRi 8-core Atom board. Two of ports (enp0s20f2 & enp0s20f2) are LACP bonded to form bond0. I am getting packet drop notifications on all enp0s20f2, enp0s20f2 & bond0 at different intervals (at least one email a day).

The main switch is a Ubiquiti UniFi US-24, again with aggregation configured.

The main communication from this server is to the FreeNAS server on the same switch. The NAS uses the same 8-core Atom board, and also with LACP configured in the same way.

Everything SEEMS normal to me with my general usage.

So my main question is - should I be worried? I guess I should?

And what/where should I start looking where the problem is? Unfortunately I have no clue, but am keen to learn.

Is the problem Netdata reporting too aggressively or incorrectly? NICs on the Proxmox or FreeNAS malfunctioning? The switch not aggregating correctly? Mis-configuration on the Proxmox or FreeNAS?

In the past I had a D-Link DGS-1210-28P and I didn't have as many packet drops as I have now. But the D-Link died a premature death.

Thanks.



No comments:

Post a Comment