External metrics server

[warllo]

Hi,

I have the following config on my proxmox server however metrics do not appear as reported.

Proxmox Config
influxdb:
server 192.168.2.147
port 8089

Influx db config

[[udp]]
enabled = true
bind-address = "0.0.0.0:8089"
database = "proxmox"
batch-size = 1000
batch-timeout = "1s"
retention-policy = ""

Running: tcpdump udp -i vmbr0 -vv port 8089 yields tcpdump: listening on vmbr0, link-type EN10MB (Ethernet), capture size 262144 bytes and no results. Doesn't look like proxmox is sending any data.

Any suggestions?


Thanks

 

[LnxBil]

Hmm ... I have no solution for your problem, but it should work if you just put it in /etc/pve/status.cfg with the correct indent. It was kinda super easy, just enter save and voila. Influxdb does need a restart.

Could you monitor with tcpdump that the packages are actually leaving your node? What about firewalling? Are you running the current version of PVE?

 

[warllo]

I am running Proxmox Version 5.2-10.

I could use some help interpreting the result of the tcp dump, here are the results:

root@vmh1:~# tcpdump udp -i vmbr0 -vv port 8089
tcpdump: listening on vmbr0, link-type EN10MB (Ethernet), capture size 262144 bytes
^C
0 packets captured
1 packet received by filter
0 packets dropped by kernel


I don't have the firewall running on the Proxmox host nor the influxdb server at this time.

I'm really baffled by this.

Thanks for the response @LnxBil

 

[Rob Loan]

I can offer my working config:

in /etc/pve/status.cfg

influxdb:
  server 149.59.225.132
  port 8089

in /etc/influxdb/influxdb.conf

[[udp]]
enabled = true
database = "proxmox"

 

[warllo]

Thank you @Rob Loan.

Your config resolved the issue. Appears I had some extra options on the influxdb side of things.

While this is now working. One of my hosts will only report every 4-8 minutes where as the other one will report every 15-30 seconds.

Anyone have any guesses? I have restarted the affected host.

 

[CSV]

could be a udp bufferring issue on the influxdb server. If you have a lot of ves/vms, it's going to send a lot of udp packets at once. The kernel params sysctl net.core.rmem_max and sysctl net.core.rmem_default control the size of the udp buffers. Might have to up those and/or look at influx's internal buffers. you could tcpdump from your proxmox server with the appropriate flags to show the packet contents and grep out the hosts that appear to only be every few minutes. Chances are, they're getting sent, but are getting dropped by the kernel on the influx server... or somewhere in-between (network firewalls can do this and make your day miserable)