在1.6版本中，Thrift的source和sink支持kerberos认证。客户端需要使用secureRpcClientFactory的getThriftInstance方法获得SecureThriftRpcClient对象。SecureThriftClient继承ThriftRpcClient（实现了RpcClient接口）。使用SecureRpcClientFactory依赖于Flume-ng-auth模块。客户端的principal以及keytab都需要通过参数的形式传入，他们作为kerberos KDC的证书。另外，目标服务器的principal也需要提供。下面就是secureRpclientFacotry的例子：

    ThriftSource则需要配置成kerberos模式。

    2. a1.sources = r1
    3. a1.sinks = k1
    4. a1.channels.c1.type = memory
    5. a1.sources.r1.channels = c1
    6. a1.sources.r1.type = thrift
    7. a1.sources.r1.bind = 0.0.0.0
    8. a1.sources.r1.port = 41414
    9. a1.sources.r1.kerberos = true
    10. a1.sources.r1.agent-principal = flume/server.example.org@EXAMPLE.ORG
    11. a1.sources.r1.agent-keytab = /tmp/flume.keytab
    12. a1.sinks.k1.channel = c1
    13. a1.sinks.k1.type = logger

    为了更灵活一些，failover flume client实现FailoverRpcClient，可以基于下面的配置：

    1. client.type = default_failover
    2. hosts = h1 h2 h3 # at least one is required, but 2 or
    3. # more makes better sense
    4. hosts.h2 = host2.example.org:41414
    5. hosts.h3 = host3.example.org:41414
    7. # specified, 3 in this case). A '0'
    8. # value doesn't make much sense because
    9. # it will just cause an append call to
    10. # immmediately fail. A '1' value means
    11. # that the failover client will try only
    12. # once to send the Event, and if it
    13. # fails then there will be no failover
    14. # to a second client, so this value
    15. # causes the failover client to
    16. # degenerate into just a default client.
    17. # It makes sense to set this value to at
    18. # least the number of hosts that you
    19. # specified.
    20. batch-size = 100 # Must be >=1 (default: 100)
    21. connect-timeout = 20000 # Must be >=1000 (default: 20000)
    22. request-timeout = 20000 # Must be >=1000 (default: 20000)

    Flume客户端SDK也支持在多个主机中负载均衡。client使用<host>:<port>的形式组成一个负载均衡组。client端会配置负载均衡的策略，可能是随机选择配置的主机，也可能是基于轮询的模式。你可以通过实现LoadBalancingRpcClient$HostSelector接口，指定自定义的类。在这种情况下，FQCN需要指定成特定的host selector.负载均衡RPC目前不支持Thrift.

    最大的backoff事件可以通过maxBackoff进行配置。默认是30S（在OrderSelector中指定）。backoff 参数会以指数级增长。最大限制为65536秒，即18.2小时。

    也可以直接如下配置:

    1. client.type = default_loadbalance
    3. hosts.h2 = host2.example.org:41414
    4. hosts.h3 = host3.example.org:41414
    5. backoff = false # Specifies whether the client should
    6. # back-off from (i.e. temporarily
    7. # blacklist) a failed host
    8. # (default: false).
    9. maxBackoff = 0 # Max timeout in millis that a will
    10. # remain inactive due to a previous
    11. # failure with that host (default: 0,
    12. # which effectively becomes 30000)
    13. host-selector = round_robin # The host selection strategy used
    14. # when load-balancing among hosts
    15. # (default: round_robin).
    16. # Other values are include "random"
    17. # or the FQCN of a custom class
    18. # that implements
    19. # LoadBalancingRpcClient$HostSelector
    20. batch-size = 100 # Must be >=1 (default: 100)
    21. connect-timeout = 20000 # Must be >=1000 (default: 20000)

    配置嵌入式节点与配置普通节点类似。下面是额外的配置：