DESCRIPTION

Simplify teardown

Teardown was run on setup to cleanup if tests didn't cleanup
after themselves. This isn't the right approach, test should
always end in a clean state.

Change-Id: I752dd7825507042bd11178f96d91a8c86e97e598


COMMENTS

author: Ayla Reynolds
date: 2012-02-02 15:25:19.373000000

Patch Set 1:

I didn't run the full test suite to verify, but only the failover test
test_view_failover_multiple_design_docs_x_node_replica_y,num-design-docs=20,num-docs=10000,replica=1,failover-factor=1

together with all ViewBasicTests. From the basic tests only one failed (test_view_on_20k_docs_20_design_docs) but this seems unrelated.

-------------------------------------
author: Reid Stokes
date: 2012-02-02 15:58:38.922000000

Patch Set 1: Looks good to me, approved



-------------------------------------
author: Reid Stokes
date: 2012-02-02 16:08:50.289000000

Patch Set 1: I would prefer that you didn't submit this

Suppose in cluster of 3 nodes. We are failing over node 1. As a part of failover we stop node1. Next for some reason if failover fails, the test will fail and exit out. For the next test run, the cluster is left in inconsistent state, with 2 nodes up and 1 node down. All the following tests in the run list will fail as well, bcoz one node is stopped...
I dont see how this solves the above issue of inconsistent state of the cluster.

-------------------------------------
author: Ayla Reynolds
date: 2012-02-02 16:11:00.124000000

Patch Set 1:

Alright. My fault, I ran the failover test, then the basic tests. There it didn't make a difference whether the cluster has the same number of nodes or not.

I'll try to run the tests in a way that it fails at one point without doing the undo_failover and then I'll fix it.

-------------------------------------
