rebalancer: refactoring of rebalancer_request_state

Before this patch the function `rebalancer_request_state` returned only
nil in case of errors (e.g. presence of SENDING, RECEIVING, GARBAGE
buckets, not active rebalancer). This makes it inconsistent compared to
other rebalancer functions.

Now, in case of errors we return `(nil, err)` instead of `nil`. It can
help us to propagate a meaningful error in rebalancer service.

NO_TEST=refactoring
NO_DOC=refactoring

Author: mrForza

test/rebalancer/rebalancer.result

@@ -318,9 +318,6 @@ _bucket:update({150}, {{'=', 2, vshard.consts.BUCKET.RECEIVING}})
 ---
 - [150, 'receiving']
 ...
-wait_rebalancer_state("Some buckets are not active", test_run)
----
-...
 _bucket:update({150}, {{'=', 2, vshard.consts.BUCKET.ACTIVE}})
 ---
 - [150, 'active']

test/rebalancer/rebalancer.test.lua

@@ -156,7 +156,6 @@ util.map_bucket_protection(test_run, {REPLICASET_1}, false)
 test_run:switch('box_1_a')
 vshard.storage.rebalancer_enable()
 _bucket:update({150}, {{'=', 2, vshard.consts.BUCKET.RECEIVING}})
-wait_rebalancer_state("Some buckets are not active", test_run)
 _bucket:update({150}, {{'=', 2, vshard.consts.BUCKET.ACTIVE}})
 vshard.storage.sync()
 

test/unit/rebalancer.result

@@ -290,9 +290,13 @@ build_routes(replicasets)
 vshard.storage.internal.is_master = true
 ---
 ...
-get_state = vshard.storage._rebalancer_request_state
----
-...
+get_state = function()                                                            \
+    local replicaset, active_buckets = vshard.storage._rebalancer_request_state() \
+    if replicaset == nil then                                                     \
+        active_buckets.trace = nil                                                \
+    end                                                                           \
+    return replicaset, active_buckets                                             \
+end                                                                               \
 _bucket = box.schema.create_space('_bucket')
 ---
 ...
@@ -318,13 +322,18 @@ get_state()
 ---
 - bucket_active_count: 2
   bucket_pinned_count: 0
+- null
 ...
 _bucket:replace{1, consts.BUCKET.RECEIVING}
 ---
 - [1, 'receiving']
 ...
-get_state()
+_, err = get_state()
+---
+...
+err.message
 ---
+- Replica _ has receiving buckets
 ...
 vshard.storage.internal.is_master = false
 ---

test/unit/rebalancer.test.lua

@@ -76,7 +76,13 @@ build_routes(replicasets)
 -- Test rebalancer local state.
 --
 vshard.storage.internal.is_master = true
-get_state = vshard.storage._rebalancer_request_state
+get_state = function()                                                            \
+    local replicaset, active_buckets = vshard.storage._rebalancer_request_state() \
+    if replicaset == nil then                                                     \
+        active_buckets.trace = nil                                                \
+    end                                                                           \
+    return replicaset, active_buckets                                             \
+end                                                                               \
 _bucket = box.schema.create_space('_bucket')
 pk = _bucket:create_index('pk')
 status = _bucket:create_index('status', {parts = {{2, 'string'}}, unique = false})
@@ -86,7 +92,8 @@ _bucket:replace{3, consts.BUCKET.SENT}
 get_state()
 
 _bucket:replace{1, consts.BUCKET.RECEIVING}
-get_state()
+_, err = get_state()
+err.message
 vshard.storage.internal.is_master = false
 
 assert(not vshard.storage.internal.this_replicaset)

vshard/storage/init.lua

@@ -2851,12 +2851,10 @@ local function rebalancer_service_f(service, limiter)
             return
         end
         if not status or replicasets == nil then
-            local err = status and total_bucket_active_count or replicasets
-            if err then
-                limiter:log_error(err, service:set_status_error(
-                    'Error during downloading rebalancer states: %s', err))
-            end
-            log.info('Some buckets are not active, retry rebalancing later')
+            local err = total_bucket_active_count
+            limiter:log_error(err, service:set_status_error(
+                'Error during downloading rebalancer states: %s',
+                err))
             service:set_activity('idling')
             lfiber.testcancel()
             lfiber.sleep(consts.REBALANCER_WORK_INTERVAL)
@@ -2945,18 +2943,23 @@ local function rebalancer_request_state()
         return nil, err
     end
     if not M.is_rebalancer_active or rebalancing_is_in_progress() then
-        return
+        return nil, lerror.make('Rebalancer is not active or is in progress')
     end
     local _bucket = box.space._bucket
     local status_index = _bucket.index.status
+    local repl_id = M.this_replica and M.this_replica.id or '_'
+    local buckets_invalid_state = 'Replica %s has %s buckets'
     if #status_index:select({BSENDING}, {limit = 1}) > 0 then
-        return
+        return nil, lerror.make(string.format(buckets_invalid_state, repl_id,
+                                              consts.BUCKET.SENDING))
     end
     if #status_index:select({BRECEIVING}, {limit = 1}) > 0 then
-        return
+        return nil, lerror.make(string.format(buckets_invalid_state, repl_id,
+                                              consts.BUCKET.RECEIVING))
     end
     if #status_index:select({BGARBAGE}, {limit = 1}) > 0 then
-        return
+        return nil, lerror.make(string.format(buckets_invalid_state, repl_id,
+                                              consts.BUCKET.GARBAGE))
     end
     return {
         bucket_active_count = status_index:count({BACTIVE}),