Check etcd port instead of process name

currently the pgrep check does not allow any etcd process to exist
other than the one we intend to start in our scripts. All we need
to know is if the port is free. So let us use netstat to check
if anyone is using that port. We don't really need to know if there
is another instance of etcd is running or not.
This commit is contained in:
Davanum Srinivas 2018-02-09 08:59:56 -05:00
parent aee2cff1b8
commit ea7547cd42

View File

@ -27,10 +27,10 @@ kube::etcd::validate() {
exit 1
}
# validate it is not running
if pgrep -x etcd >/dev/null 2>&1; then
kube::log::usage "etcd appears to already be running on this machine (`pgrep -xl etcd`) (or its a zombie and you need to kill its parent)."
kube::log::usage "retry after you resolve this etcd error."
# validate etcd port is free
if netstat -nat | grep "[\.:]${ETCD_PORT:?} .*LISTEN" >/dev/null 2>&1; then
kube::log::usage "unable to start etcd as port ${ETCD_PORT} is in use. please stop the process listening on this port and retry."
kube::log::usage "`netstat -nat | grep "[\.:]${ETCD_PORT:?} .*LISTEN"`"
exit 1
fi