e2e: v2 backup test

[gyuho]

Fix #5367.

[heyitsanthony]

defer os.RemoveAll(backupDir)?

[gyuho]

Just fixed. Thanks!

[gyuho]

So these are the commands that are run by this test.

./bin/etcd --name testname0 --listen-client-urls http://localhost:20000 --advertise-client-urls http://localhost:20000 --listen-peer-urls http://localhost:20001 --initial-advertise-peer-urls http://localhost:20001 --initial-cluster="testname0=http://localhost:20001" --initial-cluster-token new --data-dir testname0.etcd
./bin/etcdctl --endpoints http://localhost:20000 backup --data-dir testname0.etcd --backup-dir test-backup.etcd
./bin/etcd --name testname0 --listen-client-urls http://localhost:20000 --advertise-client-urls http://localhost:20000 --listen-peer-urls http://localhost:20001 --initial-advertise-peer-urls http://localhost:20001 --initial-cluster="testname0=http://localhost:20001" --initial-cluster-token new --data-dir test-backup.etcd --force-new-cluster

Confirmed that this command panics if I run this manually, without @xiang90's fix.

But somehow when we run on e2e, it's not passing the message type ConfChangeRemoveNode, so doesn't panic even without the Xiang's panic.

Do you see anything that I am doing wrong here?

/cc @xiang90 @heyitsanthony

Thanks!

[heyitsanthony]

can this be a prefix instead of a path so it can support multiple nodes?

[gyuho]

Right. Will fix.

[gyuho]

ah, when it starts, it deletes the data dir. That was why it wasn't panic-ing.

Will fix that too.

[gyuho]

@heyitsanthony All addressed. PTAL.

Thanks.

[heyitsanthony]

keepDataDir?

[gyuho]

fixed.

[heyitsanthony]

lgtm

[gyuho]

Thanks merging after all greens. Found some other errors.