Propagate etcd version from cluster.yaml to etcdadm #722
Conversation
k8s-ci-robot
added
the
cncf-cla: yes
Codecov Report
@@ Coverage Diff @@
## master #722 +/- ##
=======================================
Coverage 34.33% 34.33%
=======================================
Files 55 55
Lines 3073 3073
=======================================
Hits 1055 1055
Misses 1859 1859
Partials 159 159Continue to review full report at Codecov.
|
| @@ -618,6 +618,7 @@ write_files: | |||
| ETCD_ADVERTISE_CLIENT_URLS=https://$advertised_hostname:2379 | |||
| ETCD_LISTEN_PEER_URLS=https://$private_ip:2380 | |||
| ETCD_INITIAL_ADVERTISE_PEER_URLS=https://$advertised_hostname:2380" >> /var/run/coreos/etcd-environment | |||
| ETCD_VERSION={{.Etcd.Version}} | |||
There was a problem hiding this comment.
there is an easy to miss redirect happening in this file, line you add should be inside that redirect to have an effect
There was a problem hiding this comment.
not sure i get you properly, can you clarify with code example ?
There was a problem hiding this comment.
Sorry this is confusing but given what we have so far:
echo "KUBE_AWS_ASSUMED_HOSTNAME=$advertised_hostname
*snip*
ETCD_INITIAL_ADVERTISE_PEER_URLS=https://$advertised_hostname:2380" >> /var/run/coreos/etcd-environment
ETCD_VERSION={{.Etcd.Version}}
It should be the something like the below:
echo "KUBE_AWS_ASSUMED_HOSTNAME=$advertised_hostname
*snip*
ETCD_INITIAL_ADVERTISE_PEER_URLS=https://$advertised_hostname:2380
ETCD_VERSION={{.Etcd.Version}}" >> /var/run/coreos/etcd-environment
or ETCD_VERSION would be just thrown away.
There was a problem hiding this comment.
Thanks for spotting it guys. Actually first commit was both wrong and redundant, i removed it.
Cloud formation based propagation is enough to fix the problem
ytsarev
force-pushed
the
upstream/etcdversion
branch
from
3ec2d82
to
f932d51
Compare
|
@ytsarev Ah, that made sense! Thanks for your contribution 👍 |
Without this additional propagation
etcdadmfalls back to default3.1.3version as inhttps://github.com/iflix-letsplay/kube-aws/blob/master/etcdadm/etcdadm#L88
In case when any other version is specified in
cluster.yamlit will create the deviation between running etcd version and the one that is used for periodic snapshotting, e.g. running3.1.8in the main container and still using default3.1.3foretcdadmbased operations.This PR fixes the deviation.