[PATCH v2] replication: fix exit with ER_NO_SUCH_USER during bootstrap

[Serge Petrenko <sergepetrenko@tarantool.org>]

When replication is configured via some user created in box.once()
function and box.once() takes more than replication_timeout seconds
to execute, appliers recieve ER_NO_SUCH_USER error, which they don't
handle. This leads to occasional test failures in replication suite.
Fix this by handling the aforementioned case in applier_f() and add a
test case.

Closes #3637
---
https://github.com/tarantool/tarantool/issues/3637
https://github.com/tarantool/tarantool/tree/sp/gh-3637-replication-tests-fix

Changes in v2:
  - add a test and ensure new relevant
    lines are covered.
  - merge ER_NOSUCH_USER case with
    ER_ACCESS_DENIED due to similarity.

 src/box/applier.cc                      |  8 ++++++--
 test/replication/autobootstrap.result   | 19 +++++++++++++++++++
 test/replication/autobootstrap.test.lua | 10 ++++++++++
 test/replication/replica_bootstrap.lua  | 30 ++++++++++++++++++++++++++++++
 test/replication/replica_bootstrap1.lua |  1 +
 test/replication/replica_bootstrap2.lua |  1 +
 6 files changed, 67 insertions(+), 2 deletions(-)
 create mode 100644 test/replication/replica_bootstrap.lua
 create mode 120000 test/replication/replica_bootstrap1.lua
 create mode 120000 test/replication/replica_bootstrap2.lua

diff --git a/src/box/applier.cc b/src/box/applier.cc
index dbb4d05f9..02ec729bc 100644
--- a/src/box/applier.cc
+++ b/src/box/applier.cc
@@ -602,8 +602,12 @@ applier_f(va_list ap)
 				applier_log_error(applier, e);
 				applier_disconnect(applier, APPLIER_LOADING);
 				goto reconnect;
-			} else if (e->errcode() == ER_ACCESS_DENIED) {
-				/* Invalid configuration */
+			} else if (e->errcode() == ER_ACCESS_DENIED ||
+				   e->errcode() == ER_NO_SUCH_USER) {
+				/*
+				 * Invalid configuration or box.once()
+				 * hasn't finished on bootstrap master.
+				 */
 				applier_log_error(applier, e);
 				applier_disconnect(applier, APPLIER_DISCONNECTED);
 				goto reconnect;
diff --git a/test/replication/autobootstrap.result b/test/replication/autobootstrap.result
index 91badc1f1..9837aafb8 100644
--- a/test/replication/autobootstrap.result
+++ b/test/replication/autobootstrap.result
@@ -231,3 +231,22 @@ _ = test_run:cmd("switch default")
 test_run:drop_cluster(SERVERS)
 ---
 ...
+--
+-- Test case for gh-3637. Before the fix replica would exit with
+-- an error. Now check that we don't hang and successfully connect.
+--
+SERVERS = {'replica_bootstrap1', 'replica_bootstrap2'}
+---
+...
+test_run:create_cluster(SERVERS, "replication", {args="0.1"})
+---
+...
+test_run:wait_fullmesh(SERVERS)
+---
+...
+print("ok")
+---
+...
+test_run:drop_cluster(SERVERS)
+---
+...
diff --git a/test/replication/autobootstrap.test.lua b/test/replication/autobootstrap.test.lua
index 752d5f317..bbf6f6512 100644
--- a/test/replication/autobootstrap.test.lua
+++ b/test/replication/autobootstrap.test.lua
@@ -108,3 +108,13 @@ _ = test_run:cmd("switch default")
 -- Stop servers
 --
 test_run:drop_cluster(SERVERS)
+
+--
+-- Test case for gh-3637. Before the fix replica would exit with
+-- an error. Now check that we don't hang and successfully connect.
+--
+SERVERS = {'replica_bootstrap1', 'replica_bootstrap2'}
+test_run:create_cluster(SERVERS, "replication", {args="0.1"})
+test_run:wait_fullmesh(SERVERS)
+print("ok")
+test_run:drop_cluster(SERVERS)
diff --git a/test/replication/replica_bootstrap.lua b/test/replication/replica_bootstrap.lua
new file mode 100644
index 000000000..9a6c5f4a0
--- /dev/null
+++ b/test/replication/replica_bootstrap.lua
@@ -0,0 +1,30 @@
+#!/usr/bin/env tarantool
+
+local INSTANCE_ID = string.match(arg[0], "%d")
+local USER = 'cluster'
+local PASSWORD = 'pass'
+local SOCKET_DIR = require('fio').cwd()
+local TIMEOUT = tonumber(arg[1])
+local CON_TIMEOUT = arg[2] and tonumber(arg[2]) or 100 * 365 * 86400
+
+local function instance_uri(instance_id)
+    return SOCKET_DIR..'/replica_bootstrap'..instance_id..'.sock'
+end
+
+require('console').listen(os.getenv('ADMIN'))
+
+box.cfg({
+    listen=instance_uri(INSTANCE_ID),
+    replication={
+        USER..':'..PASSWORD..'@'..instance_uri(1),
+        USER..':'..PASSWORD..'@'..instance_uri(2)
+    },
+    replication_timeout=TIMEOUT,
+    replication_connect_timeout=CON_TIMEOUT
+})
+
+box.once('bootstrap', function()
+    require('fiber').sleep(3 * TIMEOUT)
+    box.schema.user.create(USER, {password=PASSWORD})
+    box.schema.user.grant(USER, 'replication')
+end)
diff --git a/test/replication/replica_bootstrap1.lua b/test/replication/replica_bootstrap1.lua
new file mode 120000
index 000000000..2f3645a72
--- /dev/null
+++ b/test/replication/replica_bootstrap1.lua
@@ -0,0 +1 @@
+replica_bootstrap.lua
\ No newline at end of file
diff --git a/test/replication/replica_bootstrap2.lua b/test/replication/replica_bootstrap2.lua
new file mode 120000
index 000000000..2f3645a72
--- /dev/null
+++ b/test/replication/replica_bootstrap2.lua
@@ -0,0 +1 @@
+replica_bootstrap.lua
\ No newline at end of file
-- 
2.15.2 (Apple Git-101.1)