删除存储桶中的对象，但存储大小未增加到riak中

Question

我设置了一个由riak，riak-cs和stanchion组成的系统到一个服务器中，我使用s3tool将对象放入riak，但当我删除对象时，它会通知“文件s3://buckets/object deleted"，我在/var/lib/riak中检查大小，然后它就不会增加。不过，我在/etc/riak/app.config文件中添加了{delete_mode，immediate}

文件配置：

[
 {riak_api, [
  {pb_backlog, 64},
    {pb, [ {"192.168.1.223", 8087 } ]}
            ]},
            {riak_core, [
            {default_bucket_props, [{allow_mult, true}]}, {ring_state_dir, "/var/lib/riak/ring"},  
              {ring_creation_size, 64},     
              {http, [ {"192.168.1.223", 8098 } ]},
              {handoff_port, 8099 },
              {dtrace_support, false},        
              {platform_bin_dir, "/usr/sbin"},
              {platform_data_dir, "/var/lib/riak"},
              {platform_etc_dir, "/etc/riak"},
              {platform_lib_dir, "/usr/lib64/riak/lib"},
              {platform_log_dir, "/var/log/riak"}
             ]},
 {riak_kv, [

         {map_js_vm_count, 0},
    {reduce_js_vm_count, 0},
    {hook_js_vm_count, 0},
                {delete_mode, immediate},  
             {add_paths, ["/usr/lib64/riak-cs/lib/riak_cs-1.5.0/ebin"]},
                {storage_backend, riak_cs_kv_multi_backend},
                {multi_backend_prefix_list, [{<<"0b:">>, be_blocks}]},
                {multi_backend_default, be_default},
                {multi_backend, [
                    {be_default, riak_kv_eleveldb_backend, [
                        {max_open_files, 50},
                        {data_root, "/var/lib/riak/leveldb"}
                    ]},
                    {be_blocks, riak_kv_bitcask_backend, [
                        {data_root, "/var/lib/riak/bitcask"}
                    ]}
                ]},    
            {anti_entropy, {on, []}},
            {anti_entropy_build_limit, {1, 3600000}}, 
            {anti_entropy_expire, 604800000},
            {anti_entropy_concurrency, 2},
            {anti_entropy_tick, 15000},      
            {anti_entropy_data_dir, "/var/lib/riak/anti_entropy"},
            {anti_entropy_leveldb_opts, [{write_buffer_size, 4194304},
                                         {max_open_files, 20}]},
            {mapred_name, "mapred"},
            {mapred_2i_pipe, true},
            {js_max_vm_mem, 8},
            {js_thread_stack, 16},
            {http_url_encoding, on},       
            {vnode_vclocks, true},
            {listkeys_backpressure, true},
            {fsm_limit, 50000}, 
            {object_format, v1}
           ]},
 {riak_search, [
                {enabled, false}
               ]},
 {merge_index, [

                {data_root, "/var/lib/riak/merge_index"},
                {buffer_rollover_size, 1048576},  
                {max_compact_segments, 20}
               ]},
 {bitcask, [   
             {io_mode, erlang},

             {data_root, "/var/lib/riak/bitcask"}
           ]},
 {eleveldb, [
             {data_root, "/var/lib/riak/leveldb"}
            ]},
 {lager, [
            {handlers, [ 
                           {lager_file_backend, [ 
                               {"/var/log/riak/error.log", error, 10485760, "$D0", 5}, 
                               {"/var/log/riak/console.log", info, 10485760, "$D0", 5} 
                           ]} 
                       ] },
            {crash_log, "/var/log/riak/crash.log"},
            {crash_log_msg_size, 65536},    
            {crash_log_size, 10485760},        
            {crash_log_date, "$D0"},
            {crash_log_count, 5},
            {error_logger_redirect, true},
            {error_logger_hwm, 100}
        ]},
 {riak_sysmon, [
         {process_limit, 30},
         {port_limit, 2},
         {gc_ms_limit, 0},
         {heap_word_limit, 40111000},
         {busy_port, true},
         {busy_dist_port, true}
        ]},
 {sasl, [
         {sasl_error_logger, false}
        ]},
 {riak_control, [
                {enabled, false},
                {auth, userlist},
                {userlist, [{"user", "pass"}
                           ]},
                {admin, true}
                ]}
].

有什么帮助吗？

干杯,

Nghiep

Answer 1

Riak CS也有一个leeway_seconds设置。其目的是保留已删除对象的数据块，以允许在delete命令之前开始下载的较慢客户端完成下载。默认设置为24小时。这意味着在Riak CS中删除的对象在24小时后才有资格进行垃圾回收。在此之后，下一次GC运行将从Riak中删除数据块，并将其替换为tombstones。如果您将delete_mode设置为immediate，则Riak tombstone将立即从后端删除。

但是，在合并操作处理包含旧值的文件之前，Bitcask后端不会释放磁盘空间。riak-admin vnode-status命令将显示每个Bitcask后端中每个数据文件的碎片百分比、死字节和总大小。