Skip to content

Commit

Permalink
Merge pull request #622 from basho/feature/confbal/schema-and-cli
Browse files Browse the repository at this point in the history 
L(-2) Config / Command-line work

Reviewed-by: seancribbs
  • Loading branch information
@borshop
borshop committed Oct 14, 2014
2 parents 86faa77 + d980e1b commit 060a6f4
Show file tree
Hide file tree
Showing 6 changed files with 943 additions and 238 deletions.
215 changes: 126 additions & 89 deletions priv/riak_repl.schema
View file
Original file line number Diff line number Diff line change
Expand Up @@ -2,144 +2,181 @@
%% Replication config

%% @doc Path (relative or absolute) to the working directory for the
%% replication process
%% replication process.
{mapping, "mdc.data_root", "riak_repl.data_root", [
{default, "{{repl_data_root}}"}
{default, "{{repl_data_root}}"},
{datatype, directory}
]}.

%% @doc The cluster manager will listen for connections from remote
%% clusters on this ip and port. Every node runs one cluster manager,
%% but only the cluster manager running on the cluster_leader will
%% service requests. This can change as nodes enter and leave the
%% cluster. The value is a combination of an IP address (**not
%% hostname**) followed by a port number
%% hostname**) followed by a port number.
{mapping, "mdc.cluster_manager", "riak_core.cluster_mgr", [
{default, {"{{cluster_manager_ip}}", {{cluster_manager_port}} }},
{datatype, ip}
]}.

%% @doc The hard limit of fullsync workers that will be running on the
%% source side of a cluster across all nodes on that cluster for a
%% fullsync to a sink cluster. This means if one has configured
%% fullsync for two different clusters, both with a
%% max_fssource_cluster of 5, 10 fullsync workers can be in
%% progress. Only affects nodes on the source cluster on which this
%% parameter is defined via the configuration file or command line
{mapping, "mdc.max_fssource_cluster", "riak_repl.max_fssource_cluster", [
%% @doc The fullsync strategy to use.
{mapping, "mdc.fullsync.strategy", "riak_repl.fullsync_strategy",
[{datatype, {enum, [keylist, aae]}},
{default, keylist},
hidden
]}.

%% @doc The limit of fullsync workers running per source-side of a
%% replication connection. For example, if you have a cluster that is
%% replicating to 3 sink clusters and this set to 5, you will have at
%% most 15 workers in total.
{mapping, "mdc.fullsync.source.max_workers_per_cluster", "riak_repl.max_fssource_cluster", [
{datatype, integer},
{default, 5}
]}.

%% @doc Limits the number of fullsync workers that will be running on
%% each individual node in a source cluster. This is a hard limit for
%% all fullsyncs enabled; additional fullsync configurations will not
%% increase the number of fullsync workers allowed to run on any node.
%% Only affects nodes on the source cluster on which this parameter is
%% defined via the configuration file or command line
{mapping, "mdc.max_fssource_node", "riak_repl.max_fssource_node", [
%% @doc The limit of fullsync workers running per node in the source
%% cluster. This setting is independent of the number of replication
%% connections. Thus, multiple simultaneous sink connections from the
%% source cluster will have to share the source node's number of
%% maximum connections. For example, if you have a cluster that is
%% replicating to 3 sink cluster and this is set to 1, you will have
%% at most 1 worker per node.
{mapping, "mdc.fullsync.source.max_workers_per_node", "riak_repl.max_fssource_node", [
{datatype, integer},
{default, 1}
]}.

%% @doc Limits the number of fullsync workers allowed to run on each
%% individual node in a sink cluster. This is a hard limit for all
%% fullsync sources interacting with the sink cluster. Thus, multiple
%% simultaneous source connections to the sink cluster will have to
%% share the sink node's number of maximum connections. Only affects
%% nodes on the sink cluster on which this parameter is defined via
%% the configuration file or command line.
{mapping, "mdc.max_fssink_node", "riak_repl.max_fssink_node", [
%% @doc The limit of fullsync workers running per node in the sink
%% cluster. This setting is independent of the number of replication
%% connections. Thus, multiple simultaneous source connections to the
%% sink cluster will have to share the sink node's number of maximum
%% connections. For example, if you have a cluster that is replicating
%% from 3 source clusters and this is set to 1, you will have at most
%% 1 worker per node.
{mapping, "mdc.fullsync.sink.max_workers_per_node", "riak_repl.max_fssink_node", [
{datatype, integer},
{default, 1}
]}.

%% @doc Whether to initiate a fullsync on initial connection from the
%% secondary cluster
{mapping, "mdc.fullsync_on_connect", "riak_repl.fullsync_on_connect", [
{datatype, {enum, [true, false]}},
{default, true}
%% @doc Whether to initiate a fullsync on initial connection from a
%% sink cluster.
{mapping, "mdc.fullsync.start_on_connect", "riak_repl.fullsync_on_connect", [
{datatype, flag},
{default, on}
]}.

%% @doc a single integer value representing the duration to wait in
%% minutes between fullsyncs, or a list of {clustername,
%% time_in_minutes} pairs for each sink participating in fullsync
%% replication.
{mapping, "mdc.fullsync_interval.$cluster_name", "riak_repl.fullsync_interval", [
{datatype, {duration, ms}},
{include_default, "all"},
{commented, "30m"}
%% @doc The duration to wait between initiating fullsyncs for all
%% connected sink clusters. If set to "never", fullsync will not be
%% automatically initiated. If set to "per_sink", individual intervals
%% should be set using mdc.fullsync.interval.<sink-cluster-name>
%% @see mdc.fullsync.interval.$cluster_name
{mapping, "mdc.fullsync.interval", "riak_repl.fullsync_interval", [
{datatype, [{duration, m}, {atom, per_sink}, {atom, never}]},
{default, "6h"}
]}.

%% @doc The duration to wait between initiating fullsync with a
%% specific connected sink cluster.
%% @see mdc.fullsync.interval
{mapping, "mdc.fullsync.interval.$cluster_name", "riak_repl.fullsync_interval", [
{datatype, {duration, m}},
{include_default, "sink_cluster"},
{commented, "2d"}
]}.

{translation,
"riak_repl.fullsync_interval",
fun(Conf) ->
Minute = fun(Millis) -> Millis div 60000 end,
FullSyncIntervals = cuttlefish_variable:filter_by_prefix("mdc.fullsync_interval", Conf),
case proplists:get_value(["mdc", "fullsync_interval", "all"], FullSyncIntervals) of
undefined ->
[ {list_to_atom(Name), Minute(Value)} || {["mdc", "fullsync_interval", Name], Value} <- FullSyncIntervals];
X -> Minute(X)
end
FullSyncIntervals = cuttlefish_variable:filter_by_prefix("mdc.fullsync.interval", Conf),
{[{_, Global}], Sinks} = lists:partition(fun({I, _}) -> ["mdc", "fullsync", "interval"] == I end, FullSyncIntervals),
if Global == never, Sinks == [] ->
disabled;
Global == never ->
cuttlefish:warn("mdc.fullsync.interval is set to never,"
" sink-specific intervals are ignored"),
disabled;
is_integer(Global), Sinks == [] ->
Global;
Global == per_sink, Sinks == [] ->
cuttlefish:invalid("Cannot set mdc.fullsync.interval = per_sink and"
" omit sink-specific intervals, set sink-specific"
" intervals or use 'never'");
Global == per_sink ->
[{list_to_atom(SinkName), Value} || {["mdc", "fullsync", "interval", SinkName], Value} <- Sinks ];
true ->
cuttlefish:invalid("Cannot set both mdc.fullsync.interval and"
" sink-specific intervals")
end
end}.

%% @doc By default, fullsync replication will try to coordinate with
%% other Riak subsystems that may be contending for the same
%% resources. This will help to prevent system response degradation
%% under times of heavy load from multiple background tasks. To
%% disable background coordination, set this parameter to off.
{mapping, "mdc.fullsync.background_manager", "riak_repl.fullsync_use_background_manager", [
{datatype, flag},
{default, off},
hidden
]}.

%% @doc How frequently the metrics for fullsync source processes should
%% be gathered. The output of `riak-repl status` is calculated on this
%% interval.
{mapping, "mdc.fullsync.source.metrics_refresh_interval", "riak_repl.fullsync_stat_refresh_interval", [
{datatype, {duration, ms}},
{commented, "1m"},
hidden
]}.

%% @doc The maximum size the realtime replication queue can grow to
%% before new objects are dropped. Defaults to 100MB. Dropped objects
%% will need to be replication with a fullsync.
{mapping, "mdc.rtq_max_bytes", "riak_repl.rtq_max_bytes", [
%% before new objects are dropped.
{mapping, "mdc.realtime.queue_max_bytes", "riak_repl.rtq_max_bytes", [
{datatype, bytesize},
{default, "100MB"}
]}.

%% @doc Enable Riak CS proxy_get and block filter.
{mapping, "mdc.proxy_get", "riak_repl.proxy_get", [
{datatype, {enum, [on, off]}},
{default, off}
%% @doc Whether heartbeats are enabled for realtime replication
%% connections.
{mapping, "mdc.realtime.heartbeat", "riak_repl.rt_heartbeat_interval", [
{datatype, flag},
{default, on}
]}.

{translation,
"riak_repl.proxy_get",
fun(Conf) ->
case cuttlefish:conf_get("mdc.proxy_get", Conf) of
on -> enabled;
off -> disabled;
_ -> disabled
end
end}.

%% @doc A heartbeat message is sent from the source to the sink every
%% heartbeat_interval. Setting heartbeat_interval to undefined
%% disables the realtime heartbeat. This feature is only available in
%% Riak Enterprise 1.3.2+.
{mapping, "mdc.realtime.heartbeat_interval", "riak_repl.rt_heartbeat_interval", [
%% @doc When heartbeats are enabled, this setting is the interval
%% between heartbeat messages over the realtime replication
%% connection.
%% @see mdc.realtime.heartbeat
{mapping, "mdc.realtime.heartbeat.interval", "riak_repl.rt_heartbeat_interval", [
{datatype, {duration, s}},
{default, "15s"}
]}.

%% @doc If a heartbeat response is not received in
%% rt_heartbeat_timeout seconds, then the source connection exits and
%% will be re-established. This feature is only available in Riak
%% Enterprise 1.3.2+.
{mapping, "mdc.realtime.heartbeat_timeout", "riak_repl.rt_heartbeat_timeout", [
%% @doc When heartbeats are enabled, this setting is the amount of
%% time to wait for a heartbeat response from the sink. If a heartbeat
%% response is not received within this time, then the source
%% connection closes and will be re-established.
%% @see mdc.realtime.heartbeat
{mapping, "mdc.realtime.heartbeat.timeout", "riak_repl.rt_heartbeat_timeout", [
{datatype, {duration, s}},
{default, "15s"}
]}.

%% @doc By default, fullsync replication will try to coordinate with other
%% riak subsystems that may be contending for the same resources. This will help
%% to prevent system response degradation under times of heavy load from multiple
%% background tasks. To disable background coordination, set this parameter to false.
%% Enterprise 2.0+.
{mapping, "mdc.fullsync.use_bg_manager", "riak_repl.fullsync_use_background_manager", [
{datatype, {enum, [true, false]}},
{level, advanced},
{default, false}
]}.
{translation, "riak_repl.rt_heartbeat_interval",
fun(Conf) ->
case cuttlefish:conf_get("mdc.realtime.heartbeat", Conf) of
true ->
cuttlefish:conf_get("mdc.realtime.heartbeat.interval", Conf);
false ->
undefined
end
end}.

%% @doc How frequently the stats for fullsync source processes should be
%% gathered. Requests for fullsync status always returned the most recently
%% gathered data, and thus can be at most as old as this value.
{mapping, "mdc.fullsync.stat_refresh_interval", "riak_repl.fullsync_stat_refresh_interval", [
{datatype, {duration, ms}},
{commented, "1m"}
%% @doc Enable Riak CS proxy_get and block filter.
{mapping, "mdc.proxy_get", "riak_repl.proxy_get", [
{datatype, {flag, {on, enabled}, {off, disabled}}},
{default, off},
hidden
]}.

Loading

0 comments on commit 060a6f4

Please sign in to comment.