Fix bugs when docker restart

src/storage/fts/ob_whitespace_ft_parser.cpp

@@ -76,6 +76,8 @@ namespace storage
       || OB_UNLIKELY(0 >= word_len)) {
     ret = OB_INVALID_ARGUMENT;
     LOG_WARN("invalid arguments", K(ret), KPC(param), KP(allocator), KP(word), K(word_len));
+  } else if (word_len < FT_MIN_WORD_LEN || word_len > FT_MAX_WORD_LEN) {
+    LOG_DEBUG("skip too small or large word", K(ret), K(word_len));
   } else if (OB_ISNULL(buf = static_cast<char *>(allocator->alloc(word_len)))) {
     ret = OB_ALLOCATE_MEMORY_FAILED;
     LOG_WARN("fail to allocate word memory", K(ret), K(word_len));

src/storage/fts/ob_whitespace_ft_parser.h

@@ -25,6 +25,9 @@ namespace storage
 
 class ObSpaceFTParser final
 {
+public:
+  static const int64_t FT_MIN_WORD_LEN = 3;
+  static const int64_t FT_MAX_WORD_LEN = 84;
 public:
   static int segment(
       lib::ObFTParserParam *param,

tools/docker/standalone/Dockerfile

@@ -5,7 +5,7 @@ ARG VERSION
 ARG STEP
 
 RUN yum install -y yum-utils && \
-    yum-config-manager --add-repo https://mirrors.aliyun.com/oceanbase/OceanBase.repo && \
+    yum-config-manager --add-repo https://mirrors.oceanbase.com/oceanbase/OceanBase.repo && \
     sed -i 's/$releasever/7/' /etc/yum.repos.d/OceanBase.repo && \
     yum install -y ob-deploy obclient ob-sysbench libaio bc libselinux-utils zip && \
     rm -rf /usr/obd/mirror/remote/* && \
@@ -36,6 +36,7 @@ COPY init_store_for_fast_start.py /root/boot/
 ENV PATH /root/boot:$PATH
 ENV LD_LIBRARY_PATH /home/admin/oceanbase/lib:/root/ob/lib:$LD_LIBRARY_PATH
 
+STOPSIGNAL SIGTERM
 WORKDIR /root
 CMD _boot
 

tools/docker/standalone/step_2_boot/_boot

@@ -122,6 +122,13 @@ function deploy_failed {
   fi
 }
 
+function cleanup() {
+  obd cluster stop $OB_CLUSTER_NAME
+  exit 0
+}
+
+trap 'cleanup' SIGTERM
+
 # We should decide whether the observer's data exists and
 # whether the obd has the information of the cluster
 
@@ -261,5 +268,7 @@ else
     echo "Please check the log file ${OB_HOME_PATH}/log/observer.log"
   fi
 fi
-exec tail -f /dev/null
+while :; do
+    sleep 1
+done
 

unittest/storage/test_fts_plugin.cpp

@@ -79,8 +79,8 @@ class ObTestAddWord final : public lib::ObFTParserParam::ObIAddWord
 {
 public:
   static const char *TEST_FULLTEXT;
-  static const int64_t TEST_WORD_COUNT = 9;
+  static const int64_t TEST_WORD_COUNT = 5;
-  static const int64_t TEST_WORD_COUNT_WITHOUT_STOPWORD = 6;
+  static const int64_t TEST_WORD_COUNT_WITHOUT_STOPWORD = 4;
 public:
   ObTestAddWord();
   virtual ~ObTestAddWord() = default;
@@ -99,8 +99,8 @@ private:
 const char *ObTestAddWord::TEST_FULLTEXT = "OceanBase fulltext search is No.1 in the world.";
 
 ObTestAddWord::ObTestAddWord()
-  : words_{"oceanbase", "fulltext", "search", "is", "no", "1", "in", "the", "world"},
+  : words_{"oceanbase", "fulltext", "search", "the", "world"},
-    words_without_stopword_{"oceanbase", "fulltext", "search", "no", "1", "world"},
+    words_without_stopword_{"oceanbase", "fulltext", "search", "world"},
     ith_word_(0)
 {
 }
@@ -451,6 +451,42 @@ TEST_F(ObTestFTParseHelper, test_parse_fulltext)
   }
 }
 
+TEST_F(ObTestFTParseHelper, test_min_and_max_word_len)
+{
+  common::ObSEArray<ObFTWord, 16> words;
+  int64_t doc_length = 0;
+
+  // word len = 2;
+  const char *word_len_2 = "ab";
+  ASSERT_EQ(OB_SUCCESS, parse_helper_.segment(cs_type_, word_len_2, std::strlen(word_len_2), doc_length, words));
+  ASSERT_EQ(0, words.count());
+
+  // word len = 3;
+  const char *word_len_3 = "abc";
+  ASSERT_EQ(OB_SUCCESS, parse_helper_.segment(cs_type_, word_len_3, std::strlen(word_len_3), doc_length, words));
+  ASSERT_EQ(1, words.count());
+
+  // word len = 4;
+  const char *word_len_4 = "abcd";
+  ASSERT_EQ(OB_SUCCESS, parse_helper_.segment(cs_type_, word_len_4, std::strlen(word_len_4), doc_length, words));
+  ASSERT_EQ(1, words.count());
+
+  // word len = 76;
+  const char *word_len_76 = "abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz";
+  ASSERT_EQ(OB_SUCCESS, parse_helper_.segment(cs_type_, word_len_76, std::strlen(word_len_76), doc_length, words));
+  ASSERT_EQ(1, words.count());
+
+  // word len = 84;
+  const char *word_len_84 = "abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz123456";
+  ASSERT_EQ(OB_SUCCESS, parse_helper_.segment(cs_type_, word_len_84, std::strlen(word_len_84), doc_length, words));
+  ASSERT_EQ(1, words.count());
+
+  // word len = 85;
+  const char *word_len_85 = "abcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyzabcdefghijklmnopqrstuvwxyz1234567";
+  ASSERT_EQ(OB_SUCCESS, parse_helper_.segment(cs_type_, word_len_85, std::strlen(word_len_85), doc_length, words));
+  ASSERT_EQ(0, words.count());
+}
+
 class ObTestNgramFTParseHelper : public ::testing::Test
 {
 public: