etcd2 wont start

[Greg Fausak]

I am trying to convert to v2. My v1 configuration for etcd v0.4 is pretty simple:

/usr/bin/etcd \
        --name 10.132.90.172 \
        --addr 10.132.90.172:4001 \
        --bind-addr 0.0.0.0 \
        --discovery https://discovery.etcd.io/094b25814f0f97567f029de61927846c \
        --data-dir /var/lib/etcd \
        --http-read-timeout 86400 \
        --peer-addr 10.132.90.172:7001 \

        --snapshot true

 Works great.

Now, trying to use 2.0.8 which I just downloaded, I can't get etcd2 to start:

/usr/bin/etcd2 --name 10.132.90.172 \
    --discovery https://discovery.etcd.io/01fd8a63115c889bbef4ddb3ad3ae5c1 \
    --advertise-client-urls 'http://localhost:2379,http://localhost:4001' \
    --data-dir  /var/lib/etcd2 --initial-advertise-peer-urls 'http://localhost:2380,http://localhost:7001' \
    --listen-client-urls 'http://localhost:2379,http://localhost:4001' \
    --listen-peer-urls 'http://localhost:2380,http://localhost:7001' \

    --force-new-cluster

2015/04/20 15:42:25 etcd: listening for peers on http://localhost:2380
2015/04/20 15:42:25 etcd: listening for peers on http://localhost:7001
2015/04/20 15:42:25 etcd: listening for client requests on http://localhost:2379
2015/04/20 15:42:25 etcd: listening for client requests on http://localhost:4001

2015/04/20 15:08:23 discovery: found self 607caf39a05d8574 in the cluster

2015/04/20 15:08:23 discovery: found 1 peer(s), waiting for 2 more

and it stays here forever, i get a timeout every now and then...

2015/04/20 15:51:21 discovery: during waiting for other nodes connection to https://discovery.etcd.io timed out, retrying in 2s
2015/04/20 15:51:23 discovery: found self 7ff5d39aedd47d3f in the cluster

2015/04/20 15:51:23 discovery: found 1 peer(s), waiting for 2 moreI tried getting rid of the --discovery (which I dearly need) 

and it still won't start. 

so i toss out the https://discovery.etcd.io/01fd8a63115c889bbef4ddb3ad3ae5c1 and add 

--initial-cluster 'default=http://localhost:2380,default=http://localhost:7001'

that gets my etcd2 service started, but, I am missing the discovery write of this node (I am using that write as an indication to the waiting minions that they can now start, it also gives the minions the ip address of the master.

-g

[Alex Crawford]

On 04/20, Greg Fausak wrote:
> 2015/04/20 15:08:23 discovery: found 1 peer(s), waiting for 2 more
> and it stays here forever, i get a timeout every now and then...

It looks like you discovery token is set for a cluster of three machines
(https://discovery.etcd.io/01fd8a63115c889bbef4ddb3ad3ae5c1/_config/size).
If you only want a cluster size of one, you'll need to use
discovery.etcd.io/new?size=1.

-Alex

[Greg Fausak]

:-)

I thought it was unbound.

https://coreos.com/docs/cluster-management/setup/cluster-discovery/

thanks,

-g

[Greg Fausak]

I successfully moved over to etcd2.

-g