示例#1
0
    def _common_test_body_failed_swap_rebalance(self):
        master = self.servers[0]
        rest = RestConnection(master)
        num_initial_servers = self.num_initial_servers
        creds = self.input.membase_settings
        intial_severs = self.servers[:num_initial_servers]

        self.log.info("CREATE BUCKET PHASE")
        SwapRebalanceBase.create_buckets(self)

        # Cluster all starting set of servers
        self.log.info("INITIAL REBALANCE PHASE")
        RebalanceHelper.rebalance_in(intial_severs, len(intial_severs) - 1)

        self.log.info("DATA LOAD PHASE")
        loaders = SwapRebalanceBase.start_load_phase(self, master)

        # Wait till load phase is over
        SwapRebalanceBase.stop_load(loaders, do_stop=False)
        self.log.info("DONE LOAD PHASE")

        # Start the swap rebalance
        current_nodes = RebalanceHelper.getOtpNodeIds(master)
        self.log.info("current nodes : {0}".format(current_nodes))
        toBeEjectedNodes = RebalanceHelper.pick_nodes(master, howmany=self.num_swap)
        optNodesIds = [node.id for node in toBeEjectedNodes]
        if self.swap_orchestrator:
            status, content = ClusterHelper.find_orchestrator(master)
            self.assertTrue(status, msg="Unable to find orchestrator: {0}:{1}".\
            format(status, content))
            # When swapping all the nodes
            if self.num_swap is len(current_nodes):
                optNodesIds.append(content)
            else:
                optNodesIds[0] = content

        for node in optNodesIds:
            self.log.info("removing node {0} and rebalance afterwards".format(node))

        new_swap_servers = self.servers[num_initial_servers:num_initial_servers + self.num_swap]
        for server in new_swap_servers:
            otpNode = rest.add_node(creds.rest_username, creds.rest_password, server.ip)
            msg = "unable to add node {0} to the cluster"
            self.assertTrue(otpNode, msg.format(server.ip))

        if self.swap_orchestrator:
            rest = RestConnection(new_swap_servers[0])
            master = new_swap_servers[0]

        self.log.info("DATA ACCESS PHASE")
        loaders = SwapRebalanceBase.start_access_phase(self, master)

        self.log.info("SWAP REBALANCE PHASE")
        rest.rebalance(otpNodes=[node.id for node in rest.node_statuses()],
            ejectedNodes=optNodesIds)

        # Rebalance is failed at 20%, 40% and 60% completion
        for i in [1, 2, 3]:
            expected_progress = 20 * i
            self.log.info("FAIL SWAP REBALANCE PHASE @ {0}".format(expected_progress))
            RestHelper(rest).rebalance_reached(expected_progress)
            bucket = rest.get_buckets()[0].name
            pid = StatsCommon.get_stats([master], bucket, "", "pid")[master]
            command = "os:cmd(\"kill -9 {0} \")".format(pid)
            self.log.info(command)
            killed = rest.diag_eval(command)
            self.log.info("killed {0}:{1}??  {2} ".format(master.ip, master.port, killed))
            BaseTestCase._wait_warmup_completed(self, [master], bucket, wait_time=600)
            time.sleep(5)

            rest.rebalance(otpNodes=[node.id for node in rest.node_statuses()],
                ejectedNodes=optNodesIds)

        self.assertTrue(rest.monitorRebalance(),
            msg="rebalance operation failed after adding node {0}".format(toBeEjectedNodes))

        # Stop loaders
        SwapRebalanceBase.stop_load(loaders)

        self.log.info("DONE DATA ACCESS PHASE")
        #for bucket in rest.get_buckets():
        #    SwapRebalanceBase.verify_data(new_swap_servers[0], bucket_data[bucket.name].get('inserted_keys'),\
        #        bucket.name, self)
        #    RebalanceHelper.wait_for_persistence(master, bucket.name)

        self.log.info("VERIFICATION PHASE")
        SwapRebalanceBase.items_verification(master, self)
示例#2
0
    def _common_test_body_failed_swap_rebalance(self):
        master = self.servers[0]
        rest = RestConnection(master)
        num_initial_servers = self.num_initial_servers
        creds = self.input.membase_settings
        intial_severs = self.servers[:num_initial_servers]

        self.log.info("CREATE BUCKET PHASE")
        SwapRebalanceBase.create_buckets(self)

        # Cluster all starting set of servers
        self.log.info("INITIAL REBALANCE PHASE")
        RebalanceHelper.rebalance_in(intial_severs, len(intial_severs) - 1)

        self.log.info("DATA LOAD PHASE")
        loaders = SwapRebalanceBase.start_load_phase(self, master)

        # Wait till load phase is over
        SwapRebalanceBase.stop_load(loaders, do_stop=False)
        self.log.info("DONE LOAD PHASE")

        # Start the swap rebalance
        current_nodes = RebalanceHelper.getOtpNodeIds(master)
        self.log.info("current nodes : {0}".format(current_nodes))
        toBeEjectedNodes = RebalanceHelper.pick_nodes(master,
                                                      howmany=self.num_swap)
        optNodesIds = [node.id for node in toBeEjectedNodes]
        if self.swap_orchestrator:
            status, content = ClusterHelper.find_orchestrator(master)
            self.assertTrue(status, msg="Unable to find orchestrator: {0}:{1}".\
            format(status, content))
            # When swapping all the nodes
            if self.num_swap is len(current_nodes):
                optNodesIds.append(content)
            else:
                optNodesIds[0] = content

        for node in optNodesIds:
            self.log.info(
                "removing node {0} and rebalance afterwards".format(node))

        new_swap_servers = self.servers[
            num_initial_servers:num_initial_servers + self.num_swap]
        for server in new_swap_servers:
            otpNode = rest.add_node(creds.rest_username, creds.rest_password,
                                    server.ip)
            msg = "unable to add node {0} to the cluster"
            self.assertTrue(otpNode, msg.format(server.ip))

        if self.swap_orchestrator:
            rest = RestConnection(new_swap_servers[0])
            master = new_swap_servers[0]

        self.log.info("DATA ACCESS PHASE")
        loaders = SwapRebalanceBase.start_access_phase(self, master)

        self.log.info("SWAP REBALANCE PHASE")
        rest.rebalance(otpNodes=[node.id for node in rest.node_statuses()],
                       ejectedNodes=optNodesIds)

        # Rebalance is failed at 20%, 40% and 60% completion
        for i in [1, 2, 3]:
            expected_progress = 20 * i
            self.log.info(
                "FAIL SWAP REBALANCE PHASE @ {0}".format(expected_progress))
            RestHelper(rest).rebalance_reached(expected_progress)
            bucket = rest.get_buckets()[0].name
            pid = StatsCommon.get_stats([master], bucket, "", "pid")[master]
            command = "os:cmd(\"kill -9 {0} \")".format(pid)
            self.log.info(command)
            killed = rest.diag_eval(command)
            self.log.info("killed {0}:{1}??  {2} ".format(
                master.ip, master.port, killed))
            BaseTestCase._wait_warmup_completed(self, [master],
                                                bucket,
                                                wait_time=600)
            time.sleep(5)

            rest.rebalance(otpNodes=[node.id for node in rest.node_statuses()],
                           ejectedNodes=optNodesIds)

        self.assertTrue(
            rest.monitorRebalance(),
            msg="rebalance operation failed after adding node {0}".format(
                toBeEjectedNodes))

        # Stop loaders
        SwapRebalanceBase.stop_load(loaders)

        self.log.info("DONE DATA ACCESS PHASE")
        #for bucket in rest.get_buckets():
        #    SwapRebalanceBase.verify_data(new_swap_servers[0], bucket_data[bucket.name].get('inserted_keys'),\
        #        bucket.name, self)
        #    RebalanceHelper.wait_for_persistence(master, bucket.name)

        self.log.info("VERIFICATION PHASE")
        SwapRebalanceBase.items_verification(master, self)