Question: how to watch for cluster events?

[celevra]

Hi,

i'm on a 3 node CoreOS and i want to listn to cluster events, i think fleet does this allready, doesn't it?

I need to know as fast as possible if one node is down, so that i can switch my ClusterIP's
etcd knows allready when a node is available or not so how can i listn to that events?
I don't want to call /health every few seconds, i think there must be a elegant way.

regards

celevra

[xiang90]

There is no magic way to do that actually. Internally, etcd also does simple health checking between every pairs of nodes. Externally, you have to ping each node to know its status against your local node since connectivity it per node not per cluster.

[xiang90]

If you are talking about actual cluster events like remove/add nodes, there is already an issue for it :#5277

[celevra]

Thanks for the answer 
If Polling for /health is the right way how fast can I do it without disturbing etcd? Every 2 seconds?

---

Von: Xiang Li [email protected]
Gesendet: 17.05.2016 4:58 vorm.
An: coreos/etcd
Cc: Philipp Zeitschel; Author
Betreff: Re: [coreos/etcd] Question: how to watch for cluster events? (#5357)

If you are talking about actual cluster events like remove/add nodes, there is already an issue for it :#5277

—
You are receiving this because you authored the thread.
Reply to this email directly or view it on GitHub