stable 1068.6.0: started to receive etcd2 overloaded messages

[Juan José Amor]

Hello all,

I recently upgraded to 1068.6.0 stable in my cluster. Since then, we receive following messages:

etcd2[1074]: failed to send out heartbeat on time (deadline exceeded for 251.432912ms)

etcd2[1074]: server is likely overloaded

also fleetctl timeout commands, which logs:

etcd2[1074]: got unexpected response error (etcdserver: request timed out)

I read this: https://github.com/coreos/etcd/issues/5154

but it is strange, since this occurs after upgrade. I'm currently testing with new hearbeat-interval and election-timeout values, but

do you know what may be happened? Do you have some other suggestion for this issue?

Thanks in advance!

Best regards,

[Xiang Li]

The logging was added later on. My best guess is that the old version etcd just did not print the warning out. But the problem existed before.