Add new end point to get disk size information for search index

Currently the only way to get the disk size information for the search
index is to use the _search_info end point. But using this point would
lead to opening the search index which is non trivial as lucene has an
overhead for opening the index.

These changes would add a new end point _search_disk_size to get the
disk size information without opening the search index. Sarnie can use
this new end point and avoid opening the search index.

BugzId:87336

6 files changed, 48 insertions, 10 deletions

diff --git a/src/clouseau_rpc.erl b/src/clouseau_rpc.erl
index 88700b2b0..ad9cb1be4 100644
--- a/src/clouseau_rpc.erl
+++ b/src/clouseau_rpc.erl
@@ -21,11 +21,14 @@
 -export([await/2, commit/2, get_update_seq/1, info/1, search/6, search/2]).
 -export([group1/7, group2/8, group2/2]).
 -export([delete/2, update/3, cleanup/1, cleanup/2]).
--export([analyze/2, version/0]).
+-export([analyze/2, version/0, disk_size/1]).
 
 open_index(Peer, Path, Analyzer) ->
     rpc({main, clouseau()}, {open, Peer, Path, Analyzer}).
 
+disk_size(Path) ->
+    rpc({main, clouseau()}, {disk_size, Path}).
+
 await(Ref, MinSeq) ->
     rpc(Ref, {await, MinSeq}).
 
diff --git a/src/dreyfus_fabric_info.erl b/src/dreyfus_fabric_info.erl
index 301e2f28a..8d3877c66 100644
--- a/src/dreyfus_fabric_info.erl
+++ b/src/dreyfus_fabric_info.erl
@@ -19,15 +19,15 @@
 -include_lib("mem3/include/mem3.hrl").
 -include_lib("couch/include/couch_db.hrl").
 
--export([go/3]).
+-export([go/4]).
 
-go(DbName, DDocId, IndexName) when is_binary(DDocId) ->
+go(DbName, DDocId, IndexName, InfoLevel) when is_binary(DDocId) ->
     {ok, DDoc} = fabric:open_doc(DbName, <<"_design/", DDocId/binary>>, []),
-    go(DbName, DDoc, IndexName);
+    go(DbName, DDoc, IndexName, InfoLevel);
 
-go(DbName, DDoc, IndexName) ->
+go(DbName, DDoc, IndexName, InfoLevel) ->
     Shards = mem3:shards(DbName),
-    Workers = fabric_util:submit_jobs(Shards, dreyfus_rpc, info, [DDoc, IndexName]),
+    Workers = fabric_util:submit_jobs(Shards, dreyfus_rpc, InfoLevel, [DDoc, IndexName]),
     RexiMon = fabric_util:create_monitors(Shards),
     Acc0 = {fabric_dict:init(Workers, nil), []},
     try
diff --git a/src/dreyfus_httpd.erl b/src/dreyfus_httpd.erl
index 4b3d8cd36..3ba61edef 100644
--- a/src/dreyfus_httpd.erl
+++ b/src/dreyfus_httpd.erl
@@ -15,7 +15,7 @@
 
 -module(dreyfus_httpd).
 
--export([handle_search_req/3, handle_info_req/3,
+-export([handle_search_req/3, handle_info_req/3, handle_disk_size_req/3,
          handle_cleanup_req/2, handle_analyze_req/1]).
 -include("dreyfus.hrl").
 -include_lib("couch/include/couch_db.hrl").
@@ -108,7 +108,7 @@ handle_search_req(Req, _Db, _DDoc, _RetryCount, _RetryPause) ->
 
 handle_info_req(#httpd{method='GET', path_parts=[_, _, _, _, IndexName]}=Req
                   ,#db{name=DbName}, #doc{id=Id}=DDoc) ->
-    case dreyfus_fabric_info:go(DbName, DDoc, IndexName) of
+    case dreyfus_fabric_info:go(DbName, DDoc, IndexName, info) of
         {ok, IndexInfoList} ->
             send_json(Req, 200, {[
                 {name,  <<Id/binary,"/",IndexName/binary>>},
@@ -122,6 +122,21 @@ handle_info_req(#httpd{path_parts=[_, _, _, _, _]}=Req, _Db, _DDoc) ->
 handle_info_req(Req, _Db, _DDoc) ->
     send_error(Req, {bad_request, "path not recognized"}).
 
+handle_disk_size_req(#httpd{method='GET', path_parts=[_, _, _, _, IndexName]}=Req, #db{name=DbName}, #doc{id=Id}=DDoc) ->
+    case dreyfus_fabric_info:go(DbName, DDoc, IndexName, disk_size) of
+        {ok, IndexInfoList} ->
+            send_json(Req, 200, {[
+                {name,  <<Id/binary,"/",IndexName/binary>>},
+                {search_index, {IndexInfoList}}
+            ]});
+        {error, Reason} ->
+            send_error(Req, Reason)
+    end;
+handle_disk_size_req(#httpd{path_parts=[_, _, _, _, _]}=Req, _Db, _DDoc) ->
+    send_method_not_allowed(Req, "GET");
+handle_disk_size_req(Req, _Db, _DDoc) ->
+    send_error(Req, {bad_request, "path not recognized"}).
+
 handle_cleanup_req(#httpd{method='POST'}=Req, #db{name=DbName}) ->
     ok = dreyfus_fabric_cleanup:go(DbName),
     send_json(Req, 202, {[{ok, true}]});
diff --git a/src/dreyfus_httpd_handlers.erl b/src/dreyfus_httpd_handlers.erl
index d7acffa27..bf2be23b1 100644
--- a/src/dreyfus_httpd_handlers.erl
+++ b/src/dreyfus_httpd_handlers.erl
@@ -25,4 +25,5 @@ db_handler(_) -> no_match.
 
 design_handler(<<"_search">>)      -> fun dreyfus_httpd:handle_search_req/3;
 design_handler(<<"_search_info">>) -> fun dreyfus_httpd:handle_info_req/3;
+design_handler(<<"_search_disk_size">>) -> fun dreyfus_httpd:handle_disk_size_req/3;
 design_handler(_) -> no_match.
diff --git a/src/dreyfus_index_manager.erl b/src/dreyfus_index_manager.erl
index ccf0676eb..257529456 100644
--- a/src/dreyfus_index_manager.erl
+++ b/src/dreyfus_index_manager.erl
@@ -23,7 +23,7 @@
 -define(BY_PID, dreyfus_by_pid).
 
 % public api.
--export([start_link/0, get_index/2]).
+-export([start_link/0, get_index/2, get_disk_size/2]).
 
 % gen_server api.
 -export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2,
@@ -38,6 +38,9 @@ start_link() ->
 get_index(DbName, Index) ->
     gen_server:call(?MODULE, {get_index, DbName, Index}, infinity).
 
+get_disk_size(DbName, Index) ->
+    gen_server:call(?MODULE, {get_disk_size, DbName, Index}, infinity).
+
 % gen_server functions.
 
 init([]) ->
@@ -61,6 +64,11 @@ handle_call({get_index, DbName, #index{sig=Sig}=Index}, From, State) ->
         {reply, {ok, ExistingPid}, State}
     end;
 
+handle_call({get_disk_size, DbName, #index{sig=Sig}=Index}, From, State) ->
+    Path = <<DbName/binary,"/",Sig/binary>>,
+    Reply = clouseau_rpc:disk_size(Path),
+    {reply, Reply, State};
+
 handle_call({open_ok, DbName, Sig, NewPid}, {OpenerPid, _}, State) ->
     link(NewPid),
     [{_, WaitList}] = ets:lookup(?BY_SIG, {DbName, Sig}),
diff --git a/src/dreyfus_rpc.erl b/src/dreyfus_rpc.erl
index b34cf72fd..97a0526a9 100644
--- a/src/dreyfus_rpc.erl
+++ b/src/dreyfus_rpc.erl
@@ -19,7 +19,7 @@
 -import(couch_query_servers, [get_os_process/1, ret_os_process/1, proc_prompt/2]).
 
 % public api.
--export([search/4, group1/4, group2/4, info/3]).
+-export([search/4, group1/4, group2/4, info/3, disk_size/3]).
 
 % private callback
 -export([call/5, info_int/3]).
@@ -90,6 +90,17 @@ info_int(DbName, DDoc, IndexName) ->
             rexi:reply(Error)
     end.
 
+disk_size(DbName, DDoc, IndexName) ->
+    erlang:put(io_priority, {interactive, DbName}),
+    check_interactive_mode(),
+    case dreyfus_index:design_doc_to_index(DDoc, IndexName) of
+        {ok, Index} ->
+            Result = dreyfus_index_manager:get_disk_size(DbName, Index),
+            rexi:reply(Result);
+        Error ->
+            rexi:reply(Error)
+    end.
+
 get_or_create_db(DbName, Options) ->
     case couch_db:open_int(DbName, Options) of
     {not_found, no_db_file} ->