[Kea-users] Monitoring a Kea cluster

[Stefan G. Weichinger]

Am 03.01.23 um 21:27 schrieb Marek Greško:
> Hello,
> 
> I was observing similar issues. I currently workarounded it by
> setting:
> 
> "service-sockets-require-all": false, "service-sockets-max-retries":
> 1000, "service-sockets-retry-wait-time": 10000
> 
> I am not sure the workaround is good enough, but I did not observe
> such problems since setting these parameters. Hope it will fix it
> forever. Without the first line kea exited when interfaces were not
> up, if I remember well, and the other two lines keeps retrying to
> listen the interfaces which will come up later. Without them kea
> ceases listening on those interfaces not being up at the moment.
> 
> I also do not understand why such tweaking is needed. If I configure
> kea to listen on an interface I expect it to respond on it any time
> it comes up. Very weird software.

I applied your suggested settings to my primary kea-server and rebooted.

Without any other intervention after that there are leases provided now, 
so it seems to work ;-)

I prefer this fix because as far as I understand I would have to add one 
"-i" option per interface in systemd-networkd-wait-online.service (as 
suggested by Oscar).

That is harder to maintain ... there might be more vlans in the future.

So I tend to use the 3 mentioned settings for now, and will add them to 
the secondary kea node as well.

Thanks!